Techman, it seems inconceivable to me. Intel put a lot of development effort in the first X-Box, thus sharing some of the burden and creating a perfect opening for a next generation design team. They would have learned from their past mistakes and would have been able to leverage their past experience.
Apparently, there are unknown variables at this point, because staying with Intel should have been a brain-dead decision.
I'm thinking out loud here, but to me, there are several things that have become absolute requirements in the video game market.
1. Ultimate graphics performance. Although it can be argued that the CPU is far less important than the graphics chip here (just look at the X-Box1 with a Celeron equivalent CPU and game graphics that rivaled all the other consoles), the CPU does need to be powerful enough to keep up with the graphics chip. The next generation of graphics is going to be mind-blowing. It will need a powerful CPU.
2. Price. Consoles usually don't succeed that much if they are introduced above $300. X-Box did not ramp in volume until it was much lower than the intro price. Eventually, all consoles end up in the $150 range. You cannot have the CPU remain 33+% the price of the system.
3. Compatibility. Sony started this, but now gamers expect it. MSFT will be disadvantaged if they do not provide compatibility with their older system. If they choose to emulate x86, they will at least need a CPU powerful enough to do so. There is software available for PPC, however, that does a reasonable job emulating x86 already. MSFT could leverage that....
4. Software development. x86 rules the roost here in terms of available development tools. PPC has a good share of tools, too, but x86 would have made porting software from the PC much easier. Many X-Box sales might have come from PC games; then again, I don't have data on that, so maybe it's not a huge advantage after all.
IMO, unless IBM agreed to give MSFT a Power5 derived CPU with at least SMT and SIMD instructions at a $20-50 price point, MSFT made the wrong choice. The G5, while good, is not such a huge advancement over x86 to switch architectures. The development work to start from scratch on a PPC platform is going to be expensive, while x86 would have had some significant reuse.
Also, don't forget the OS implications. I would have thought MSFT would want a version of Windows running X-Box2. Are they willing to make an extraordinarily expensive port of Windows to PPC, or are they changing their corporate strategy altogether by leaving Windows out of the box?
Or is all of this just a ruse to get Intel to go cheaper on the CPU?
