Apple-Intel deal - what's the fuss?
If you read technology industry news with any regularity, chances are you’re going to be hearing a lot over the next few days about Apple’s announcement that they will begin using Intel processors in Macintosh computers, switching away from the PowerPC processors manufactured by IBM. (Apple, Motorola, and IBM joined forces in the early 1990s to design and manufacture the PowerPC chips.) There’s a lot of gossip and misinformation about what this really means, but here are the highlights as I understand them. These are in no particular order.
Apple achieves speed parity with Windows manufacturers. This is probably one of the best reasons for the switch. Apple has long had to combat the perception that its computers were slower (when in fact they were often faster) because the clock speeds at which PowerPC chips run are often lower than their Intel counterparts. Apple called this the “megahertz myth” and did their best to educate users to the fact that what actually mattered were the number of instructions a processor could run per second. Unfortunately the perception remained and the megahertz myth became a gigahertz gap, especially when IBM couldn’t produce PowerPC G5 chips at the clock speeds they claimed they would be able to, on the timetable they predicted—to the point that the fastest Macs actually were slower than the fastest Windows machines. (How they compared in actual efficiency of the operating system and software performance, I don’t know.) Perhaps more importantly, IBM has yet to produce a G5 that runs at a temperature low enough to be installed in a laptop computer. With the switch to Intel chips, Apple can match any competitor’s chip speeds.
In theory, an Intel Mac could be dual-booted into Windows. This is perhaps the most exciting aspect of the switch for me, and I suspect for most users once they really think about it. Previously, if you wanted to run Windows on a Mac, you were stuck with the awful Virtual PC emulator to do so. With an Intel processor in the box, there’s no reason that Apple couldn’t allow for Windows to also be installed to run natively on the computer. You couldn’t run them side-by-side of course, and there are some definite technical and financial reasons Apple might not want to allow such a thing. If they did, however, it would be a lot easier than it is now. There are already versions of Linux that run on Apple hardware, so it’s not as if no one’s ever thought about doing this before.
Buying a Mac will be a tricky proposition for the next year or so. Personally I was hoping for some new hardware announcements, particularly in the notebook arena, but that didn’t happen. Now I’m left wondering how long I’ll be able to stick it out without a new laptop, and what PowerPC products we’ll see announced between now and the time the new Intel Macs begin shipping. I think there will be a large number of holdouts—people will obviously want to delay their new computer purchases until the new machines appear. There’s no reason to believe, however, that the next operating system (announced as “Leopard,” if you’re the kind of person who digs the jungle cat monikers) and even the one after that won’t run on PowerPC chips. The question is whether third-party vendors will always offer two versions of their software. I’m not even sure that will be necessary—will there be “fat binary” applications that automatically detect which architecture they’re using?
For the most part, end users won’t notice the difference. Honestly. Do you notice when your car dealership starts selling cars with Goodyear tires instead of Firestone? This is just as much of a non-issue to the average Mac user. Some people will care, some people will notice, but overall it’s a lot of bitching and moaning about nothing.
5 Comments:
Where this probably will effect end users is that in some theoretical future Mac applications will come out that won't run on G-series computers. Given that Macs have longer life-spans than Wintels, this will be an issue for some people.
Seeing the rabid Mac fanboy reaction has been interesting. Hating Intel seems to be some sort of religious calling. People are swearing off Mac forever, saying they've "lost it," whatever "it" was.
I suspect the reason Apple switched was that IBM didn't seem to be making heat dissicipation a priority. As IBM is making chips for game consoles, which can have big fans, Apple is getting spanked waiting for a G5 chip they can put in a laptop without it catching on fire. Add that the laptop market is getting to be more important than the desktop market, and you have a reason for desperate action.
You'll notice I said "most" users. Given that the vast majority of people use their computers for little more than word processing, e-mail, and web surfing, I think it's a pretty safe bet.
It appears I may be wrong. Though nailing down an exact answer is still eluding me, I'm reading somethings that indicate that programs written for Macintel would still run on the G-series so long as they use "universal binaries." Maybe a seperate compiling is required, but maybe that isn't a problem.
Ugh. "Macintel" and "Mactel" are awful little contractions, aren't they?
I've seen confirmation of this on one of the O'Reilly blogs: fat binary applications will exist.
Jobs said that during the keynote: XCode will be able to generate fat binaries with a single checkbox, which presumably everyone will check. If you're in that environment, at least, this shouldn't be a big problem.
The developers who are worrying now are the ones who don't use XCode; not everything will be trivial to port. Some C/C++ code, especially written by inexperienced programmers, makes hidden assumptions about the endianness of the processor it's on, and that stuff will have to be patched.
There's other fine print. While it's easy to recompile things, it might be more work to optimize the code for the new processor. That will, I guarantee you, create situations in the first year or two in which Mac apps ported to Intel run slower than their Windows counterparts, and that will look very, very bad in benchmark comparisons: oh, no, Macs are slower even with the same processor!
Rosetta has restrictions too. It doesn't emulate Altivec, so the few high-end apps that absolutely need Altivec won't run, and the more intelligently written ones that can be optionally accelerated by Altivec will take another speed hit under emulation. Classic will not run on Intel Macs, so your old Classic apps are toast when you switch.
But the vast mass of potential switchers from Windows probably don't care about most of these things. They'll be buying low-end Macs and happily using the stuff that ships with them, regardless of the processor inside. The only thing inhibiting them might be the twelve months of confusing FUD before the Intel machines come out.
Post a Comment
<< Home