The Cell
I have been following the seepage of information about the Cell processor for a while now.
This seepage has now turned into something of a flow the last couple of days, due to a bunch of details released at the International Solid State Circuit Conference.
For the uninitiated, the Cell processor is a new cpu made by IBM, Sony and Toshiba.
It's the processor that is going to power the Playstation 3.
It's also an interesting new processor/system design.
In the beginning the information was diagrams and descriptions from patents filed by IBM.
Now there are much more understandable articles.
If you are interested in processors or computer system architectures, I'd recommend taking a look at Arstechnicas Cell two recent articles:
First Cell article
Second Cell article
If you are not technically inclined and you still want to know what it says, I'll sum it up:
These processors, assuming they will deliver what is described, will blow the good old PC-with-a-graphics-card machines away when it comes to graphics (and other things as well).
Hmm.
Interesting.
I want one.
Problem is that like any other blue sky technology, the Cell will require massive industry backing to fulfill the role given to it. Certainly, it will be an interesting piece of technology, and it will be used in many devices, but I find it hard to believe in the idea of a live network, continually discovering new CPUs and dispatching workloads to them.
Essentially they describe a parallel computer, which requires either a complete reeducation of developers, or software technology to leverage the hardware in a way that shields developers. I doubt either will happen, and I think developers will lament the PS3's method of software in the same manner as they have the PS2's.
While I agree with the fundamental observation, I am not so keen to write it off as "will never happen".
If IBM and Sony don't constitute "massive industry backing" to you, what does?.
According Sony are hiring people to work with their development tools for the PS3. They are very aware of the problems the rather strange PS2 architecture caused, and that they need to give people a better set of tools to develop for it.
But even if you end up having to program the PS3 with COBOL, I still think it's going to be the industry leader; No matter how much the developers whine, the customers are key – and the customers will want what the PS3 can deliver.