The Nuclear Option
With news that Apple is switching its Macintosh computers to Intel chips appearing in the New York Times and the Wall Street Journal, it seems like it is time to consider the possibility that this is more than a bad rumor flashback. I've spent the last day and a half driving across the desert, without an internet connection, getting myself all worked up about this.
There are a couple of possibilities here:
1) the rumor is just plain wrong. Maybe Apple is trying to fuck with the rumor sites/ferret out a leak. This would be a highly questionable business practice, however, and if the stock moved on the info, it would likely get management in deep shit. (Although with Chris Cox helming the SEC, maybe not).
2) the rumor is about using non-CPU Intel components (WiMax, or other) in Apple products. This isn't a big deal, except as to what these components bring to the platform.
3) Apple has hired Intel to produce PPC chips. This would be so fucking cool. We'd have three PPC chipmakers, instead of two! Some have speculated that Apple has the rights to a shitload of IBM PPC IP and could contract intel to make the chips more or less as-is. Another intriguing possibility (I don't know how feasible), is that Intel could graft a PPC front-end onto the vaguely RISCish innards of its Pentiumish chips. I'd imagine the innards of the various Pentium lines are sufficiently un-RISC as to make this a potentiall hairy undertaking. This is the best possible outcome, assuming the rumour is true.
4) Apple will use Intel x86 chips in its Macs. Gag me with a spoon. This would a really big fucking deal. Most developers would have to recompile their apps. Most cutting edge video and audio stuff would have to be completely re-written. Apple would almost certainly need to provide emulators for PPC-->x86 and x86-->PPC. Getting PPC emulated, in software, on x86 is gonna be a fucking nightmare. Emulating a register-based ISA on a stack-based ISA is enough to make grown men cry. Seriously. If Apple goes this route, I'll have to either abandon a current project, or take a crash course in x86 assembly. Parrot's looking pretty good in this light, isn't it??
This would require quite a bit of under-the-hood work by Apple. They have a lot of ISA specific code (mostly vector stuff, but also simple byte-order optimizations and tricks) that would need to get rewritten at some point, otherwise x86 performance is gonna suck. Contrary to popular opinion, I think most Carbon apps should be very easy to get running on x86. Pure Cocoa apps should work without modification.
,p>It is also possible that Apple will restrict x86 chips to certain products (the rumored iTablet, a set top box [but Motorola would be the logical supplier for this], or x86 xServes for nervous IT types and Apps with heavy x86 assembly code-bases. Or maybe nifty new iPods?5) Apple will announce that MacOS X will now run on COTS wintel boxes. This would be the biggest fucking deal, if it were available tomorrow. With Longhorn far off in the future, it seems to me now would be the best time to do this. This would be a big fuck-you to Microsoft and would be the ballsiest move in tech history. This is unlikely, but if there ever were a time to do this, now is it. I think this would make me happy, but I'm not sure it would be a good business move. It would have to be coupled with a big push in the office-type app space. Signs point to no.
6) ??? Got me. Something no one has thought of.
In any event, I'm writing this a block away from Moscone West. Hopefully, we should all know after tomorrow's WWDC keynote what the fuck is going on. I had thought WWDC 2005 would be booooring. Apparently not.