Intel inside

Today’s big news, for me at least, is that Apple will be moving to use Intel chips in Macs, in place of the PowerPC chip that’s been used for the last ten years or so. This isn’t good news for Mac users, since the required transitions are always messy and painful. The reason for the shift is that IBM, which produces the PowerPC, has been unwilling or unable to produce a low-heat version of the G5 chip for use in Powerbooks. The shift marks the end of the PowerPC strategy, which began as an alliance between Apple, IBM and Motorola and seemed at first likely to produce a serious alternative to Intel’s dominance of the CPU market.

The good news for Mac users is that, thanks to the massive success of the iPod and the flow-on effects to Mac sales, Apple is in a stronger position to make this move than at any time in years. In addition, the Mac OS itself is easily portable. Still, I expect some of my favorite obscure applications will struggle to make the change.

Another item in the positive column is that this ought to make it easier to run Windows on the occasions I need it (I currently use the Virtual PC emulator which is impressive, considering, but still problematic), and may also reduce the difficulty of porting Windows software to Mac.

23 thoughts on “Intel inside

  1. I don’t think you will notice the difference. OSX 10 is really just linux and that runs fine on Intel and AMD. Best of all you may be able to use the AMD Athlon 64 with the linux 64 bit extensions.

  2. OSX isn’t Linux. It’s a Mach Kernel with BSD bits. But more importantly, it’s completely different from Linux above the command-line level.

    Customers won’t notice the difference, probably. Developers sure will. The Mac is a modern computer platform; Intel X86 is medieval. Good-bye Open Firmware, good-bye Altivec. Good-bye stuff “just working”.

  3. Nick, unless you’re coding in assembler, what do you care how medieval the instruction set is?

    If Apple have done their job properly, converting apps that don’t talk directly to hardware) should largely be a matter of taking the source code and recompiling.

    Programs that write “binary” to disk files may run into a classic software porting problem called “endianness” (roughly, what order you write things), but it’s a very well-known problem and there are standard checks and fixes for it.

    Applications using Altivec directly will also have to rewrite their code to avoid it, but that will be a tiny part of a tiny subset of applications.

    And, presumably, Apple have already written their PowerPC emulation code, so all the old programs will continue to run as they did before.

    As for porting from Windows to Mac, it’ll make bugger-all difference. Most of the work in porting is dealing with the different operating systems, not the different processors.

    The other nice thing about this, from Apple’s perspective, is that into the future they have *two* sources they can buy processors from – Intel and AMD – and therefore will be in a position to drive a harder bargain on CPUs. While it’s not quite that easy, as Intel and AMD CPUs aren’t drop-in replacements for each other, every time they do a significant redesign they can choose who to go with and the software will just work.

  4. “As for porting from Windows to Mac, it’ll make bugger-all difference. Most of the work in porting is dealing with the different operating systems, not the different processors.”

    True, but at least the Mac won’t need an Intel emulator to run Windows apps, it can (presumably) run natively on the Intel CPU.

    As a developer of Windows applications, anything that allows Macs to run Windows apps would be a huge benefit to Apple. We have looked at doing a Mac version countless times, but the economics simply don’t add up; a port is hugely expensive and the Mac market is still less than 1/10th the size of the Windows market.

  5. Davidm, note I said “porting”, not emulating.

    There is an argument to suggest that being too compatible with the Windows is actually counterproductive for an alternative operating system; if the emulation works too well nobody buys software specifcially written for it. Witness OS/2.

    Frankly, I haven’t had a closed-source productivity app installed on a home machine in years, so I don’t give a toss (Windows does make a better gaming OS than Linux at this stage). If Apple were in a draconian market position they’d play even less fair than Microsoft does.

  6. I expect that this transition will be more or less transparent to me as a user and technician. I don’t recall many problems with the 68000-PPC transition 10 years ago (except performance). They announced a technology branded “Rosetta” that will allow existing PPC applications to run on the Intel hardware. The idea of Windows running in a Classic-like environment is rather appealing.

    We haven’t heard many details yet about what Apple’s going to do, but they have said “”We will not allow running Mac OS X on anything other than an Apple Mac” so the “things just working” is likely to continue.

    The big question for me is how they will prevent OS X from running on, say, a Dell. The thousands of extremely clever hackers out there will doubtless find a way to make it work on a commodity PC, and some enterprising white box manufacturer will start producing pseudo-clones. That’s a medium-run danger and I wonder how Apple will avoid it without resorting to some sort of trusted computing/hardware DRM model.

    Which, now I think about it, could be part of their master plan. Apple’s heavily involved in music, and Jobs owns a movie studio. Will they throw their weight behind a hardware DRM system?

  7. Nick said: Good-bye Open Firmware, good-bye Altivec. Good-bye stuff “just working�.

    I have no inside information on the subject, but I would bet a lot that Apple will keep Open Firmware, and a closed hardware platform (which is what makes Apples ‘just work’).

    Altivec is handled by the compiler (99% of the time, unless you are hand coding assembler), and gcc can vectorize for SSE too. Plus I hope Apple will use Intel’s compilers in place of gcc, which optimises better than gcc does.

    It won’t be tough for developers — you can download xcode 2.1 today, and compile a fat binary for PPC/ia32, although you’ll have nothing to run it on!

    The downsides are the danger that people will start running OS X on non-apple hardware by circumventing whatever mechanisms Apple use to lock people to their hardware, hurting Apple’s hardware sales and possibly the loss of 64bitness.

  8. You’re all missing the point – the big market effects aren’t about Apple but about Intel, who just got a monopoly. IBM/Motorola won’t find it worthwhile pouring billions into ever faster commodity processors – they’ll concentrate on high-margin niches now.

    So this decision by Apple will quickly be irreversible – no alternative processors will be available to either MS or Apple (Linux development for other processors will soon slow too).

    From this distance it looks like a major strategic mistake by Apple (hardly their first). Intel will have them by the balls forever now, plus there’s still the risks of cloning pointed to by others.

  9. x86 got a monopoly (although ironically the next generation games consoles are using PPC derivatives!), not Intel.

    AMD and VIA make x86 compatible CPUs.

  10. Aw, come on Tom, Via is nothing (specialised low performance ultralow power niche only) & I reckon Intel is already working hard on expanding the x86 instruction set to force AMD to play catchup again.

    So maybe Intel aint a *pure* monopoly, but it already has an awful lot of market power & this decision by Apple substantially increased it. My point about irreversibility stands – once Apple goes to Intel there’s no way out of their loving embrace.

  11. And in any case, these developments are all about the desktop. I think you’ll find that Motorola and IBM will continue to pour billions into commodity server/embedded/etc. processors (no, these are not ‘niches’) for many years to come.

  12. I’m an Australian at the conference.

    some quick points.

    * Rosetta is slow, expected to get faster.
    * Classic won’t work on Mac/x86
    * The Mac/x86 boxes are P4 3.6 Ghz with crappy Intel integrated graphics chipsets. It felt as real as any other Mac I’ve ever used… 🙂
    * The room was amazing when the announcement happened… imagine 4000 stunned mullets… we’d all heard the rumours, but hadn’t quite believed them.
    * The prevailing mood is one of optimism.
    * The Mathematica guys are good value.

  13. You’d be surprised how many people hand code for Altivec.

    The bigger problem is big companies who don’t use Xcode, but use Metroworks instead. They currently have no way of building fat binaries without refactoring their entire workflow….

  14. Derrida,

    AMD have just under 17% of the x86 market and continue to take share from Intel, so you don’t need to be too concerned about Intel’s monopoly. Apple already has the problem of IBM not taking their needs seriously enough, and Moto lost interest in Apple years ago. (IBM’s chip sales to Apple is about 2% of their fabrication division’s revenue, and I think the number is lower for Moto).

    Intel have had Apple by the balls for at least the last 7 or 8 years. Intel produce chips that run faster than PowerPC, which makes Apple product look bad. The AIM alliance (Apple/IBM/Moto aka PowerPC) hasn’t lived up to it’s early promise, and Apple’s jumped to the best performing architecture.

    Strategic error, maybe but I don’t think so. (not licencing OSA to Microsoft in the early 90’s was I think the mother of all Apple errors, but that’s another story). This has been “Plan B” at Apple for at least the last 5 years, and while I agree that it’s risky, it’s less risky than selling kit that is both slower and more expensive than the competition.

    Ab’s already mentioned this, but you’re way overstating the significance of desktop processor sales. It’s about 2% of the total chip market. Most of the remaining 98% is for embedded systems (there’s about 250 million 8-bit chips sold per month), but there’s also some action at the high end which is also where the high margins are. That’s not terribly relevant to Apple’s situation, but it is to Linux. I don’t think this will have any appreciable effect on Linux development – except maybe Linux for PowerPC.

  15. Apple move to Intel…

    So it’s official. Apple are moving to Intel processors.

    The atmosphere in the SteveNote was hilarious. He came out and said it, and… well imagine a room full of 4000 stunned mullets… OMG as the kids say. If you’re watching the keynote stream, lo…

  16. Apple move to Intel…

    So it’s official. Apple are moving to Intel processors.

    The atmosphere in the SteveNote was hilarious. He came out and said it, and… well imagine a room full of 4000 stunned mullets… OMG as the kids say. If you’re watching the keynote stream, lo…

  17. I cant see how this makes a difference to mac owners, I bought an iBook because it was like linux that works out of the box for desktop use. OSX and Apple design are major selling points. The hdd crashed on my iBook recently and I took it to the “Genius Desk” at the Apple store on Reston.

    I got asked what operating system was on it, “OSX?” I then got asked what version. I didnt know. On the iBook I use firefox, xchat, bash and emacs – what do I care what point version of the operating system I use? Who cares? That is not what I use computers for.

    For people like me; PPC, Intel, AMD – who cares, as long as it works, has OSX, excellent battery uptime and is well designed, I really dont care. What hardware they use to achieve that just isnt my concern as a consumer.

  18. I tend to agree with Cameron. Like any other Mac user, the switch to Intel is somewhat of a shock to me, but I can’t see it being a bad thing. I buy Macs for the quality hardware and software combination – the chip architecture doesn’t come in to it at all.

  19. I guess consumers should be concerned about the extent to which your PowerPC Mac will be supported in the future (eg. in relation to future OS releases). The resale value of your PPC machine just took a serious hit, too.

  20. Apple has always been good about backward compatibility. I guess
    the trouble here is the amount of energy which will be dedicated to making sure that I can run the OS built for an Intel chip on my current machine.

  21. If you want the ultimate in (lack of) computer power, go for a one instruction set chip (OISC) like the subtract and branch if negative machine. I won’t recommend the move machine as that comes too close to being practical. Google on “linux coffee machine” for more details.

Comments are closed.