OSX - the real hardware mover and shaker

Software development is always substantially slower than hardware development these days, again because the level reached by now is "sufficiently good" (that is, tolerable for the customer rather than awe-inspiringly great). This is one area where Apple is substantially better than MS - the upgrades over the last three years (OS8/9 as well as OSX) are impressive, in frequency as well as in actual improvements - speed and stability, rather than just new features.

Of course, this has been imperative as a compensation for the lack of hardware improvement, but it has also been very expensive for Apple (s/w development takes more manpower than h/w, for the same increase in speed). Now that it will actually be possible for Apple to "throw power cycles at the software code" to gain speed and compensate for code bulging, they will do so.

Cutting down on further optimization will also push users into upgrading soonest - and it is my opinion that we will see Apple return to the more speedy h/w upgrade cycles of two years ago: when the OS code keeps developing bulk and new features that has to be hardware-supported (to wit, Quartz Extreme), people won't hold back three or four years before upgrading. But customers need to be able to point at concrete improvements that they "need" in order to justify upgrading; what with the hold on substantial CPU and mobo architectural improvements available to Apple, this has not been the case. Look also at FireWire 2, GigaWire and AirPort 2 which has not happened at the speed hoped for and projected by Apple; the only possible attempts at offering "necessary" improvements have been the SuperDrive and the flat screens, together with the emphasis on aesthetics.

The Motorola PPC7457 and the rest of the Gx family for desktop computers won't be substantially upgraded from now onwards, other than a die shrink and cache changes. These can be done with little research (ie, investment) and will enable Motorola to continue improving processor speed, albeit slowly, over the next year. They will "trickle down" the ladder, reaching the iBooks by the end of next year - by which time they will be sufficiently cool (temperature-wise) and power-lean for that to happen.

IBM's upcoming PPC970, the Great White Hope of computer infighting, won't be a major upset for Intel/AMD when it shows up in October/November - by that time it will be roughly on par with the latest x86es, power wise. The good news is that not only its number-crunching but (more importantly) the whole bus architecture will be so substantially upgraded that the rest of the Apple mobos will also be on par.

The better news is that from then on (barring any substantial changes in h/w configurability on the market (for instance, a major improvement/price slide for Itanium III)), Apple will be able to keep up with (albeit conservatively) PC architectural improvements, and IBM will be able to scale the processor frequency faster than xx86es and roughly reach par with them (thus making the overall processor power available a good deal better than Intels).

Whether they - Apple in particular - will do that, is more doubtful, for reasons of economy. Improvements are almost never introduced on the market unless there is a need for it: pressure from competitors (Apple doesn't have any primary competitors, which is a major reason for the slowness of their h/w plans), or simply that it has become cheaper, or equally cheap, to use the improved version. Production is motivated by market forces, not by idealism (barring the Newton, but see what happened to it!). That also explains the evolutionary, rather than the revolutionary, changes in consumer computing.

The interesting part is that Apple has made OSX into "the next insanely great thing" that can push the new generation of hardware: the role that spreadsheets and desktop publishing has had was not taken over by video editing, as Jobs expected - it has been the operating system itself, ably supported by the iApps. That is why the OS upgrades will continue to come thick and fast, but their emphasis will move from optimization to new features that make new demands for new hardware. And with the PPC970 ability to empower new architectural and connectivity features (from October/November onwards), it will be push rather than shove.