Wired article on nexgen consoles
The next generation is looking interesting for a few different reasons. All 3 are using the same Power instruction set. Two of them are using graphics chips from the same company. At least two of them are also using Multi-core versions of the Power chips.
Xbox has been stated to have a 3 core chip with all 3 cores sharing the same memory and cache. This is similar to traditional SMP systems, except that 3 is a rather odd number for an SMP system. They have to be doing something different. Preliminary specs
Nintendo really doesn't have many specs available for their system. Of the three it seems to be the most secretive. We know that it uses an ATI GPU and a Power CPU. We know it will play gamecube games. Past that we don't really have much information.
Sony PS3 has a 9-core chip with a single core having direct access to the cache and using main memory. The other 8 cores have indirect access to main memory and have their own local memory for their respective programs. They're supposedly miniature Power cores with capabilities similar to the older PowerPCs used in the early powermacs. The primary core is more modern and will probably have multiple pipelines etc.
Depending on the capabilities of the secondary 2 cores on the Xbox compared to the limited SPE cores on the PS3, it could be a pretty even matchup on graphics. the PS3 has a definite advantage over at least the Xbox, because the Xbox will be using a mere dual-layer 7gb DVD compared to PS3's 50gb Bluray disc. Xbox will have the advantage people are claiming that Sony used last time. It will be first to market of the generation. If you recall though. The first system in this generation was not the PS2, but the Dreamcast...
Back to the original article for a moment.
"I think they will define a new standard," said Marco Spitoni, senior lead artist at Milestone, an Italian game studio. "I'm afraid many developers will not be able to make these updates, to make these leaps to next-gen consoles because of technological limitations and financial limitations."
Peter Hajba, a developer from Finland's Remedy Entertainment, agreed.
"I think content teams for them need to be huge," Hajba said. "The budgets for them need to be huge, so (I think) fewer and fewer games may be going to come out."
Hmm. An interesting point. I think we've seen the opposite though. Far more games have come out for the PS than the SNES before it. And it seems that even more games are coming out for the PS2. Especially toward the end of life for each console.
One thing about the 2 of the new systems though. They are both shooting for an easier API. Sony learned its lesson last time and is embracing existing PC API's and extending them for their own purpose. Xbox has been refining its API for awhile and is upgrading it to Direct X 10.
PS2 and Xbox level games are quite acceptable to most people. The key to the new systems is that they make that level of graphics childsplay. Someone did a demonstration where they took a number of old Atari 2600 games and reimplemented them in PyGame in under 8 hours each with one developer where they took months to make on the old systems. Games that look as good as our current generation would be much easier to make. No more low level tweaking with the vector processors needed. Lots of unused capacity so you can work with higher level languages and constructs. Content is easier because the PS3 supports more standard formats like the Nvidia CG format for high level graphical effects. The extra capacity laying about could mean that one team could develop an engine or library that could be reused by other teams more easily than in the current generation. The engines can be more abstract.
Making the top-notch titles that squeeze every shred of graphics power out of the systems will take more money and people, sure. But does every title need to be top-notch? Especially when it'll get to the point where the improvements are so small as to be unnoticable to most people.