MikeB said: why is everything going to happen in the future for the ps3. that;s all I ever read I mean it was launched november 2006 thats over 18 months ago and yet we still hear about the untapped potential of it. A lot of it is happening right now, top launch games like Motorstorm and Resistance tapped only 10-15% of the performance what can be delivered by the Cell's SPUs (this could even have been dragged down further through extensive optimisation though). Motorstorm 2 and Resistance 2 will tap a lot more of this performance. This transition is happening because the PS3's design is based on radically more modern and different technology. Things which run on the Cell's PPE nowadays will have to be split and moved over to the SPUs (which results into huge performance gains), once that's accomplished there are endless oppertunities for optimisations and enhancements. performance so great that even now they still can't run GTAIV at the same framerate or resolution as the Xbox360. That was a game with millions of dollars spent on both ports to get it right. if I was a sceptic I would be wondering how much of this is just spin There's no spin with regard to SPU potential, there are enough research documents to back up games developer claims. However these research studies are able to concentrate on specific tasks. The amount of code and functions within a game engine are huge, devs just can't do it all at once assuming their game engine is suitable for extensive multi-threading at all (if not it's better to start over from scratch, as future PCs will also see more and more cores added). The first multi-CPU consumer desktop computer was released in 1995, called the BeBox (inspired by how the very asynchronously operating Amiga computer handled multimedia back in the 80s). Sadly PC technology took until 2005 to advance towards similar technology advancements (and Windows, MacOS and Linux all handle this far less efficiently than BeOS did even back then). Dude smoke the crack less please? Seriously do I doubt that a pc in 1995 with multiple cpus is more efficient than a multicore cpu today with billions spent on optimization? For instance a research team can solely concentrate on Artificial Intelligence and show the Cell is suitable for this use like no other consumer CPU currently available, however AI is just one part of a game engine and it may make more sense to first get the physics engine onto the SPUs, which is very beneficial for many games or have audio processing done by the Cell's SPUs which is relatively an easy thing to do, etc. This is doubtful. The SPUs are fast but they lack branch prediction and a good AI script has a lot of branches. A multicore X86 cpu would be faster and for a bonus they do out of order execution as well. |
Tease.







