The Transition to x86
During yesterday’s announcement, Apple suggested the migration from PowerPC to x86 environments would be a simple, straight-forward exercise that requires little effort on the part of Macintosh software developers. Pardon us if we don’t agree with this view. Developers must translate most software from PowerPC to x86 binary code. These two architectures couldn't be more different. (One tidbit: PowerPC addresses characters in memory from left to right, like this sentence, but x86 tfel ot thgir morf sretcarahc serots.
Steve Jobs and Paul Otellini, photo by PCWeb web-site
Compilers cannot always sort out problems like these, so humans must manually inspect hundreds of code modules during the port.) Apple hopes that a software emulation package it calls “Rosetta” will permit users to run old, unmodified PowerPC code on new x86 platforms, but emulation has had a mostly checkered past.
Apple may be relying on a new emulation package (QuickTransit) from Transitive Technology, a UK-based startup. Transitive makes impressive claims regarding the performance of its product, but has only recently begun shipping a production version of its software to selected Silicon Graphics customers. Both the translation and emulation approaches require extensive testing to validate proper software operation. Even when the underlying architecture does not change, ISV’s must still qualify the operation of software on platforms with new processors or peripherals; the change in instruction set architecture seriously complicates this problem. It helps somewhat that Apple is ready to ship x86 development platforms running its OS X almost immediately, but this does not mitigate the extensive work its development partners must undertake.
It should not be unreasonable to anticipate some slowdown in PowerPC Macintosh shipments above and beyond the normal seasonal factors over the next few quarters. Given today's announcement, most Mac users will defer system purchases until the new x86 platforms arrive. Nobody wants to buy the last PowerPC-based system. Macintosh users with immediate requirements will have no choice but to buy those soon to be obsolete platforms, but others with the flexibility to time their purchases will surely delay as long as possible. Any new PowerPC platforms introduced during this period will likely be met with a tepid market response.
Today’s announcement does contain several positive elements. The shift to Intel architecture surely will be the last such transition the company has to make. There just are not that many alternatives left in any event. Even if Apple does not share the rest of the Wintel market's Windows vision, its hardware requirements have far more in common with those of Dell, HP and other Wintel OEMs than they do with Cisco, Ford, PlayStation, and other PowerPC OEMs. It made sense for Intel to invest hundreds of millions of dollars in a processor (Centrino) aimed specifically at notebook computers, given that it could sell fifty or sixty million of these chips per year. How could IBM or Freescale ever justify this kind of expense, just to satisfy Apple's need for one or two million chips per year?
Apple’s key ISVs, Microsoft and Adobe stood on the stage with Apple and pledged their support for this new CPU strategy. Had either of these companies withheld that support, Apple's plan could easily have become a non-starter.
The move opens up two new potential markets for Apple. Since the company won’t do anything to preclude these new platforms from running Windows (even if Apple has no plan to market them with Windows), it should be possible for adventurous users, or even adventurous OEMs to market the Apple x86 platforms with Microsoft's OS. While such a combination might not appeal to the masses, it would certainly be of interest to those who value the elegance of Apple’s industrial design but who cannot stray from the Windows software model. Conversely, once Macintosh software runs on x86-based industry standard platform, a “Mac Clone” market may emerge. Apple’s last half-hearted attempt at cloning during the early days of the PowerPC transition, ended in disaster and litigation.
During a Q&A session following yesterday's announcement, Phil Schiller, an Apple VP, said Apple “will not allow running Mac OS X on anything other than an Apple Mac.” Insight 64 suggests Mr. Schiller and Apple’s attorneys review section 1 of the Sherman Act, 15 U.S.C. @ 1 (1976) and section 3 of the Clayton Act, 15 U.S.C. @ 14 (1976) before they finalize their views on this matter.