Almost a week since the Apple on intel shocker, and I'm still chewing it over. There's loads I could say about it (or rather, my thoughts on it), and I guess some might filter through in posts over the coming weeks, but for starters:
- Why not Cell? I've read all the theories, and understand that timing of cell might have been less than perfect, but here's an article from itjungle.com back in December that says "IBM and Sony are developing a workstation based on Cell chips, which is the first product that IBM will ship based on the product". What's that gonna run? - Linux, I'll bet. It'll be high-end and specialised, but the fundamental principle is that you can build a workstation around Cell, and it will almost certainly kick the shit out of anything on the market at the moment in terms of pure processing grunt (more here).
- Why not AMD? - they're faster, cheaper and cooler than intel right now.
- Why not stick with IBM PPC? - AIX UNIX runs on multi-core IBM PPC servers already (although they do run very hot).
Ultimately, I think that PPC was, as Steve Jobs said, not able to provide for Apple's plans. This is probably partly technical and partly commercial, but mostly true. In terms of Cell, I think the timing is off (not available for too long, and unproven), and in terms of AMD, well, who knows. One theory is that Steve Jobs is pissed at their CEO for some old Motorola grudge.
I am almost certain, however, that there is more to come on this story; that the move to intel is part of a larger play by Apple. This might be to do with hardware DRM provision by intel to allow for movie distribution along the lines of the Apple Music Store, or it might be a more extensive plan to grow Apple into a place where OS X can really threaten Windows market share in some areas.
Or..... (and bear with me, as I'm not a developer or even a coder; I'm just a systems engineer), it could be something even bolder. I haven't read anyone else with this theory, and I'm sure that there are many reasons that I could be spouting crap, but it occurs to me that Apple have just given all their developers a means by which they can port their OS X applications from PPC to intel very easily. Like, in a few weeks or something. Furthermore, they have given them the tools to produce binaries that will run on OS X on either processor type. Seems to me that Apple have therefore just started re-writing the rules on architecture changes.
I don't really know where this leads, but I do know that there are no success stories in the annals of the computer industry regarding computer hardware companies marketing systems based on more than one architecture (although Sun are presently having a bash with Sparc and Opteron processors in their range, and Linux is doing great and runs on almost every architecture there is). Anyway, if architecture changes are more trivial (i.e. you can switch about as fast as you can ramp up production), then surely the rules change. The processor manufacturer has to work harder to keep you as a customer, as you can leave for someone else a bit more easily.
For Apple, maybe this means that they can push intel to do new and great things for them (I hope so). Or maybe the "universal binaries" that the developers will be producing will get even more options - perhaps there will be a "Sparc" checkbox in the future alongside the PPC and intel ones, placing OS X alongside Linux as a multi-platform O/S. Or maybe the universal binary will allow further separation of professional and consumer products in the future, with Cell (PPC) in Power Macs and intel in portables and home computers.
As I said, I just don't know how any of this plays out, but I think that maybe the change is at least as important as the end architecture, and based on my experience as a Mac user, those guys in Cupertino are pretty smart and pretty good at impressing me with products and ideas.
I'm gonna stick with Apple, and see what they have up their sleeve........