Apple moving to Intel^H^H^H^H^H ARM ?

The press is full of theorising about Apple moving to Intel [www.theregister.co.uk] with consequent analyst-friendly bull:

Intel’s Centrino processor, a chip based on the Pentium III core which dynamically regulates its clock speed from 300Mhz up according to system demand, finally gives Apple an alternative.

Thing is, I go with the final paragraph of this article – Intel now sells the XScale, otherwise known to UK geeks as the ARM.

An Acorn Risc Machine running MacOS – how cool would that be? Low power, too. Probably quite small/thin. Hmmm. New PDA-based product range, anyone?

Comments

9 responses to “Apple moving to Intel^H^H^H^H^H ARM ?”

  1. Stephen Usher
    re: Apple moving to Intel^H^H^H^H^H ARM ?

    Well, seeing as Apple holds enough of the rights to the PowerPC to be able to get others to build them for them, it may be that you’ll see an Intel PowerPC instruction set compatible chip. After all, Intel’s not doing very well in the Itanium or x86/x86_64 fields at the moment. ARM is faring better because there are few embedded processor competitors in that space other than Motorola.. or at least the spin-off.

    Of course, the problem Apple will have with any of the Intel architectures (unless they resurect Alpha, which Intel owns the rights to now) is that they’re all little-endian. Apple stuff has been big-endian since the original Macintosh. This will pose some major headaches, especially for application writers who will suddenly find all the code which was written assuming big-endian-ness in their code and have to re-write it, and also for any PowerPC emulator. Remember, emulating the 680×0 on a PowerPC was far simpler as the OS layer didn’t have to translate data structures, merely pass them on. When you have to translate data structures etc. on the fly you hit a huge performance drop.

  2. alecm
    ARM Endinaness?

    Is the ARM a BE or LE architecture ?

  3. acb
    re: Apple moving to Intel^H^H^H^H^H ARM ?

    But are there powerful enough ARM-architecture chips for the high end of the performance spectrum? Offering low-power Mac Micros to go with people’s iPods is one matter; keeping the architecture sufficiently scalable to allow the same software to run (with natural performance differences, of course) on both them and the high-end Mac towers is another.

  4. alecm
    re: Apple moving to Intel^H^H^H^H^H ARM ?

    Well, the rumous just say variations on a theme of:

    “Apple! Intel! Ooh! Apple! Apple! Intel! Ooh! Intel! Apple! Intel! Ooh! Ooh!”

    …and not much else, so I’d bet a small amount upon a low-end Apple device, bigger than an iPod, smaller than an iBook, sometime in the future, with an ARM in it.

    But only a small amount.

  5. Stephen Usher
    re: ARM Endinaness?

    Erm, ARM’s instruction set was based upon the 6502.

    I’m trying to remember which endian that was.. I know it was opposite to the Z80.. So it’s probably big-endian, as the Z80 is a fixed 8080, which is a forrunner of the 8086/8088 and hence x86.

    I’ve just looked it up.. I was wrong.. the 6502’s little-endian. I was obviously thinking of the 6809.

    Have a look at mindprod.com/jgloss/endian.html near the end.

  6. Stephen Usher
    re: ARM Endinaness?

    Actually, they have one bit incorrect, the DEC Alpha is agnostic, like the PowerPC, and can run in either big or little endian modes.

  7. Mart
    re: ARM Endinaness?

    ARM cores are also endian agnostic.

  8. alecm
    re: Apple moving to Intel

    Well, I was flat-out wrong about that one. Should have based my theory on observations of other major hardware manufacturers (ahem) – but then Jobs has always been one to think different… 😎

  9. Stephen Usher
    re: Apple moving to Intel

    But he *DID* think different…

    Most other hardware manufacturers are moving slightly away from Intel and going to AMD… Jobs is doing the opposite!

Leave a Reply

Your email address will not be published. Required fields are marked *