With only a few hours before the WWDC keynote, I want to be sure I take the opportunity to make some largely uninformed comments on the MacOS on Intel rumors. So many others have, why shouldn't I?
The biggest problem with a switch to Intel would seem to be the Osborne effect. John Gruber mentions this in his I'll See You Intel post. I think this is a big issue. I know that I would postpone my purchases of new Apple hardware pending the availability of Intel-based updates. On the other hand, when Intel-based hardware arrived, I would be even more likely to purchase than before.
That said, the ordering of the Intel rollout that has been reported seems a bit confusing. Reportedly Apple would start at the low end of their line (the Mac mini) and work up. That doesn't make much sense. Buyers of the Mac mini are clearly not driven by performance, and the G4 is a pretty good chip for that platform. It seems to me that the place that Intel could add the most value immediately would be in the portable line. It seems that IBM has been unable to bring the G5's power consumption and heat generation down enough to fit comfortably into an Apple notebook and the G4 chips currently in use don't seem to be scaling very well. Meanwhile, Intel's Pentium M has gotten consistently faster since its introduction and currently tops out at 2.13 GHz. Unlike the Pentium no-M, the Pentium M seems to be a rough match for the G4 in clock speed comparisons, and the G4 currently tops out at 1.67 GHz. Intel reportedly has a 64-bit Pentium M waiting in the wings. An Intel-based Apple laptop could be a good way for Apple to speed up their laptop line a bit and for Intel to showcase a 64-bit, low-power chip.
The incremental speed increase wouldn't be the only reason or even the main reason that I'd be excited to see an Intel-based Apple laptop. More than anything, I'd like a sleek, powerful laptop like my Powerbook that could also run Windows applications when necessary. Now, I doubt Apple is going to certify Windows to run on their hardware, and I don't really think I'd want to run Windows directly on the hardware anyway. I do, however, think that VMWare could have a MacOS port running pretty quickly and that would be just perfect for me. (While they're at it, I wouldn't mind having them virtualize the hardware so that I could run MacOS in a VM, as well.) In his 5 Misconceptions... post, Rui Carmo says he doubts Microsoft would go for it. I don't know why they would care, and I don't know how they would have any influence. I doubt they like VMWare allowing Windows to run on Linux, but it does.
Any processor transition like this would hinge on some sort of emulation to allow existing binaries to run on the new hardware unchanged. Apple has done this before during their transition from 680x0 to PowerPC. This transition will be somewhat different for two reasons. On the negative side, PowerPC chips were appreciably faster than 680x0 chips, and that speed difference helped offset the cost of emulation. The same isn't true for Intel chips as compared with PowerPC. On the positive side, the classic MacOS in place during the PowerPC transition still had parts of the OS itself that were still runing 680x0 code. In contrast, MacOS X will certainly run entirely natively on any new Intel hardware. That would leave only application code running under emulation, and presumably improvements in emulation technology such as that from Transitive would help mitigate performance issues there. (Interestingly, Transitive's chairman is former NeXT COO, Peter van Cuylenburg.) Such emulation should only be an interim necessity, as recompilation of MacOS X applications to run on Intel should be relatively straightforward—at least for XCode apps.
Ok, so those are some fairly meandering comments. What do I predict will happen? I have no idea, but I sure can't wait to find out. I would clearly love to have an Intel-based Powerbook, but my needs are rarely typical so I don't think that should factor into any prediction. Marketing this transition would seem quite difficult—almost impossible. Technically, I have no doubt it could be done and really without that much trouble, but I just don't see how to spin it PR-wise. That said, I think Gruber's observation in his post that it is unlikely that CNet and the WSJ would be so completely wrong in their reporting is pretty compelling. It is one thing for this to come from AppleInsider, et al., but quite another for it to come from the Wall Street Journal.
Only an hour to go. I guess I better go ahead and post, or this will be totally pointless. Probably more after the keynote.
Comments