What's Known About the PS3

1up has an expansive piece up exploring everything they know about the PlayStation 3. They cover rumours, prices, technology, and the limited information currently out there on upcoming games. From the article: "While the hard facts are still tough to nail down, the general consensus is that the PlayStation 3 is the most powerful of the three next-generation systems, although probably not by as much of a margin as Sony would like us to think. The arguments for the technical strengths of the PS3 go into CPU floating-point capabilities and the difficulties surrounding programming for parallel architectures, but the long and short of it is that whether or not the advantages of the PS3 are apparent will depend on developers' ability to utilize the PlayStation 3's unique architecture."

Comments

It's what's inside the box that counts

[wuffalicious]

I was under the impression that the original Playstation 2 took a while to come out as well, and did just fine on release despite the dreamcast being available long before it (and, in some ways, being a superior piece of hardware). I look at all the hype that's being slung around, and I wonder why people obsess so much about the systems themselves. Ultimately it's going to come down to a matter of games - it doesn't matter what your system is capable of if no one developes for it. The Dreamcast should be proof positive of that.

Release titles are what will matter - how many people would have picked up an xbox had it not been for Halo?

Re:Too many variables, too little information

[garcia]

Very few people are about to spend $800 for a console - there's a lot better ways to spend that money (you could even buy a 360, and several games).

Umm, that's how much is costs Sony, not the consumer. Sony is going to be selling it at a considerably larger loss than Microsoft is for the 360. The PS3 will likely be priced at (or very near) that of the 360.

More like "What is not known about the PS3"

[javaxman]

Really, the article is a bit more like "here's the stuff we really don't know the answer to :"

- when will it ship
- what will it cost
- will games actually be able to live up to the great graphics and model/AI processing promised by the Cell processor marketing, or will they look pretty much like XBox360 games ?
- will there be a halfway decent online component ?
- is a hard drive included ? An add-on ? What's the deal ?

All we *really* seem to know is that there's a Cell processor inside, it'll support HD, include a Blu-ray drive, will take some sort of hard drive ( at least as an add-on ), will have built-in networking ( like crazy ), and will have a *ton* of games written for it... seriously, that's a long, long list of games. Oh, and it'll play existing PS2 games, though the article doesn't say that I think it's a well-known given. That and the controller they showed just looks weird.

That's without the PS3's NVidia GPU

[Animats]

I went to a talk at Stanford where the lead architect of the Cell processor described that demo. That's running on a Cell processor which is actually doing the rendering. Sony's original plan for the PS3 worked that way, but they eventually put in a conventional NVidia chip.

So in the PS3, the Cell processors aren't doing the rendering. The Cell should render about as well as everything else with a current NVidia part.

Flyovers are easy if you have enough RAM and a GPU. How much RAM did the demo rig have?

Re:More emphasis on functional languages.

[cgibbard]

Firstly, Haskell isn't strict, it's lazy. Secondly, this isn't funny.

Code written in languages with strong algebraic properties like referential transparency is ideal for doing automatic high-level transformations of code in order to increase parallelism.

As architectures get more complicated with multiple processors and multiple pipelines, we will more and more want to rely on automatic tools to search for good ways to structure code, from breaking up major processes right down to instruction level scheduling. Using high-level information about the task at hand and the components which make it up, which isn't present in lower level languages will be important in this task.

As a precursor to this, have a look at FFTW, the Fastest Fourier Transform in the West. On the surface, it appears to be written in C, but that C code is not entirely written by hand, it is machine generated by an O'Caml program with some very specific high-level knowledge of the problem (applying some mathematics to do a directed search for an implementation of an FFT of any size which is fast on the given platform). Basically, it's using high-level properties of the problem in order to obtain very fast code implementing a solution. The more information available to a compiler, the better.

Haskell itself already provides higher level information about the overall structure of a computation, leaving more of the details to the compiler than say, your average C++ or Java program. The implementations aren't totally killer yet, but there's a *lot* of untapped potential there.

(Even now, GHC is placing first and second on the computer language shootout with default settings.)

Haskell itself isn't quite ideal for high-level machine transformation of code, but I'd contend that it's certainly a practical starting point, and it's certainly my favourite programming language to actually get things done in.

  - Cale