OK, besides a Mandelbrot viewer - I am curious what you guys would like to see as far as PS3 homebrew tutorials go.
I don't have a PS3 to play with
so I'm not sure what's happening so far. What I think would be good would be...
1) Some visual high performance application - The fractal viewer was one. It's always good to get something running with tangible results. The fractal viewer might not be ideal as a Cell wow factor application because it fits conventional CPUs well too. Perhaps something that makes use of larger data sets to leverage the LS capacity? Maybe a raytraced renderer of a fractal, which should still fit into LS but look more impressive than the typical 2D variety?
2) Useful basics - There's really a division here between two fields; games and applications. I think providing the back-bone basics for people to build off is necessary to kick-start homebrew. For the games, some form of rasterizing engine would be nice to see. Showing how such could be developed on one SPE, and then multithreaded across several, would be great. For applications, perhaps a real-time movie player with effects, showing decompression and applying some 2D effects, or even 3D transformations. That could be scaled to handling multiple video feeds and blending between them. Another application that would be quite unique would be a software synthesizer.
What I'm thinking would be ideal is a longer term project, covered in a step by step development. eg. Perhaps pick a 3D Pong game and cover rasterization of vectors, then lighting, then texturing. Or pick a movie player on a SPE, multithread it, and in part 3 add some effects. Don't run these as a tutorial either. By that, I mean don't have a multithreaded 6 channel movie player with effects finsihed and then show how it works. Give everyone the work in practice. Show them how the job is tackled with a blank page beginning that coders are faced with, and include all the bugs and workarounds and redesigns that include the thought processes. As people have to learn Cell, they would be helped by seeing how others are learning Cell and adapting their designs.
I think keeping the results very visual (or audible!) is important to generate interest. It's all very well having numbers to show, but people are more impressed by fancy visuals! Hence the original Cell demos at E3 '05. Things like the above I think would have lots of scope for the community to embelish and develop their own techniques. In my case, I wouldn't care to develop a multithreaded video player, but I would be very interested in creation special FX. Given a basic engine, I'd be sure to spend many an hour coding effects and trying out different techniques. Likewise you'll get others more interested in developing audio or animation techniques, and giving them a test platform to work from means the community isn't wasting time with individuals all creating their own basic application before they get to play.
