2005-06-23

Comments on Mac PPC → Intel transition

As promised. Most of the issues involved in the transition have been hashed out by others (probably better than I could), so I'll try not to go over that ground again.

Nonetheless, here's my summary of informed opinion: In general, the consensus seems to be, rightly, that the transition to Intel is a good business decision for Apple, made at a time of strength, which will better position Apple to concentrate on what it does best: providing a superior overall user experience (hardware + software). Intel's strength in chipset engineering is critical here (freeing Apple from this task) and, along with their industry leading fab capacity, makes them a superior partner for Apple than AMD; Apple should make a good partner for Intel since Apple's closed hardware provides a perfect outlet through which to introduce new PC technologies/standards into the marketplace. The software transition is likely to be fairly smooth for Mac-only software developers, most of whom are already using Apple's development tools; this is important because many of these developers have limited resources even as they are perhaps the most significant contributors, besides Apple itself, to the Mac user experience. Major developers like Adobe and Microsoft who aren't will have a harder time, but they have already committed to porting their apps, and will want to do so promptly in order to avoid becoming the next Quark. A significant question mark hangs over scientific and media applications that rely heavily on AltiVec; in some cases there may be no way to move their applications to x86 without a significant loss in performance, and so they will need to consider other alternatives. Hopefully Intel will work to improve the SIMD capabilities of their processors in the future. For most apps, performance on Intel is likely to be equivalent or better than on PPC.

Something which has not been emphasized enough, I think, is that the real challenge for Apple in this transition lies in execution. Too much attention has been focused on the reasons behind the transition (probably since it wasn't expected). In principle, this transition should be easier than the original transition to PPC and the later transition to OS X. Apple lost developers and market-share in each of those transitions. If that is to be avoided this time around —and given the Mac's market-share today, it had better be— Apple needs to make this transition as smooth as possible in practice. A number of things raise concern in this regard:
First, it appears that this transition was decided on very recently and, while Apple has had the foundation for such a transition (OS X running on Intel) for the past five years, many critical decisions are just now being made or have yet to be made. Offering an explicit roadmap to the transition would be a strong sign that Apple has a clear game-plan and would do a lot to build confidence in the Mac community, as well as providing something to plan around for those to may be adversely affected by the transition.
Second, following in Apple's obsessive tradition of corporate secrecy, no specific commitment has been made to address the just-mentioned deficiency This is an exceptional situation in which Apple needs to be exceptionally open and communicative, yet they are proceeding as if it's business as usual —bad sign.
Third, the high cost of the developer transition systems announced at WWDC puts them out of reach of many smaller developers which, as noted above, are major contributors to what makes the Mac user experience generally so positive; Apple needs to make sure that these developers will be able to tweak and test their apps before Intel-based Macs hit the market.
Forth, Apple cannot afford to transfer resources from existing projects to this transition. Tiger is probably the buggiest major release of OS X since 10.1, and both it and many of Apple's apps are in need of updates and improvements —Apple cannot allow development to languish further in a competitive marketplace while they concentrate on the transition.
Apple is not a company known for its ability to execute smoothly; it needs to now more than ever. Hopefully Intel, which has a better track record in this regard, will be a positive force here, both through what will surely be their heavy involvement in the design of Intel-based Macs and through their own desired to be associated with a successful transition.

One interesting matter of note is that the triumph of x86 architecture is another —and impressive— example of market failure. Despite the fact that the PPC is probably a better architecture in the abstract, Apple has little choice but to switch. It is simply not feasible for almost any computer manufacturer to fund their own processor development on an "as you go" basis: the cost are simply too high, as revealed by the disappearance of the PA-RISC, Alpha, and MIPS architectures from the market. (Sun still soldiers on with SPARC, but that architecture is dead too if their recent big bets don't pay off; specialists like Cray and the embedded market are different cases, of course). In a competitive marketplace it is important minimize costs which are not also incurred by your competitors and which do not make a sufficiently strong contribution to the success of your products: processor development is just such a cost. Ultimately we will pay a price for this loss of diversity, competition and innovation in the market for microprocessors.

That's about all I have to say on the matter for now. Whatever else might be said, the bottom line is this:
What else are Mac-folk going to do, use Windows?!?

0 Comments:

Post a Comment

Links to this post:

Create a Link

<< Home