Intel+1

Just over a year ago, Apple announced that it was going to transition the Macintosh from PowerPC processors to Intel processors. Macworld has a couple of articles on the state of the transition one year out.

The transition is going really well- probably as well, or better, than most people would have imagined as a best case scenario a year ago.

A fundamental reason for the success of the transition lies in the design of Mac OS X. The low level guts of the operating system are relatively processor agnostic, and Apple had been enforcing that neutrality (and maintaining an Intel version internally) for years. On top of its core, most of OS X's system and application services function at a relatively high level of abstraction from the hardware that runs beneath them. This means that significant changes under the hood are less likely to have serious consequences for the software that runs on top. And application recompiles to account for the new architecture are relatively straightforward.

An interesting side effect of this abstraction comes in the form of performance penalties. There's no doubt that, subjectively speaking, Mac OS X and most of the programs that run on it have always felt a little slower than Windows. There is a lot of controversy over what specifically causes this phenomenon (and whether it exists at all). But I think that Mac OS X's fundamental design, in which much of the system and application services are running at higher levels of abstraction from the nuts and bolts of the OS kernel, trades off a bit of performance in exchange for added flexibility and development convenience.

I think this tradeoff is well worth it.

In addition to having an operating system that was designed to handle a transition of this sort, Apple has managed the process quite well. In fact, the technical merits of the OS actually belong under this category, since the entire OS strategy was designed to provide this level of flexibility. From the moment the processor transition was announced, Apple has provided excellent tools to enable developers to port their code, and has over-delivered on their timing commitments for releasing the Intel machines.

The application transition strategy- Universal Binaries, which by design work on both PowerPC and Intel processors- is working well. In most cases, developers and users don't have to deal with Intel and PowerPC versions of programs- there's just one version, which works on both. This has the benefit of hiding the transition complexity from people who don't need to worry about it, while at the same time ensuring that developers continue to release software that is compatible with PowerPC systems. PowerPC users (myself included, for the time being) don't feel abandoned.

It's too bad that Adobe Creative Suite, and Microsoft Office, aren't Universal Binaries yet, and likely won't be until some point well into 2007. But at least those programs run (albeit with a speed penalty), thanks to the Rosetta binary translation layer in the Intel version of Mac OS X. This compatibility escape hatch, while not a permanent solution, is another example of good planning and execution on Apple's behalf.

Stepping back from the transition itself, it's interesting to look at a few other issues that we now have the ability to gauge. After about a decade of theoretical debate, it seems pretty clear that Intel processors are fundamentally better suited for desktop computing than PowerPC processors. This is an oversimplification, of course, but the bottom line is that the Intel Macs (running native software) are simply faster than PowerPC Macs- sometimes dramatically so. Of course, we're comparing the very latest, dual-core Intel processors to PowerPC chips that are at this point getting long in the tooth, but still- Intel seems faster.

It's also interesting to see how joining the "mainstream" has helped Apple, as opposed to hurting it. The Intel-based Macs are, from a basic hardware perspective, nearly identical to traditional PCs. This is the culmination in a trend of convergence that has been happening for years, and many have been concerned that this sort of convergence would eliminate the Mac's uniqueness and dampen its capabilities.

To the contrary, Intel Macs are more capable than ever. The intense competition between Intel and AMD means that we'll be seeing much more rapid innovation from Intel than we had been accustomed to seeing in the niche PowerPC platform. And Windows support will in all likelihood bring new users to the Mac, and will bring Macs into places they've never been before.

There may be some downside to this. Do game developers have a strong incentive to dedicate significant development resources to creating Mac versions of their games, when the Windows versions would run, unaltered, on any Mac that can run Windows? (I'm asking you, Jason.) What about other Windows developers who might have been considering a Mac release, or dual-platform vendors who can just barely justify the expense of creating an OS X version- will they continue to dedicate resources to Mac development?

I certainly hope so, and I'm optimistic. And like many Mac aficionados, I'm looking forward to Apple's Worldwide Developers Conference this August, where we'll finally get to see a preview of what's coming in Mac OS X 10.5. This is going to be a big release for Apple- more than ever, people are paying attention to the Mac. Microsoft is gearing up to release Windows Vista in the same timeframe, which will up the ante further. The Intel transition opens up some interesting possibilities with regard to Mac vs. Windows competition, now that they can theoretically play with each other as equals on the same machine.

I suspect we'll see a number of new capabilities in 10.5 that are designed to compete with Vista on a feature list perspective. I think we'll probably see some user interface design evolution, designed to make Vista seem a bit outdated even before it's released. It's quite possible that we'll see dramatic new levels of Windows compatibility. Of course, if history is a guide, the biggest "wow" will probably come in the form of something no one has really considered yet.

For the past year, Apple has been successfully re-engineering the Mac platform to enhance its abilities to be a Mac and to enable it to compete with traditional PCs in ways previously not possible. In the coming year, I hope we'll see a big push to leverage this transition in ways that raise the bar for everyone.

Along those lines, I think Apple needs to make a couple of commitments in areas where the Mac has potential vulnerabilities. First, Apple should designate a "quality czar" (I'm sure they could come up with a clever job title) whose job is to ensure that all of Apple's hardware products are ruggedly designed, thoroughly reliable, and free of absurd QA problems. This person needs to get up on stage with Steve Jobs at hardware introductions and talk for five minutes about how impressive a particular machine's testing phase was, perform some audience-pleasing durability demos... and lose his or her job if Apple's hardware QA does not improve.

Second, Apple needs to advance the issue of software security. At this moment in time, Mac OS X is far more secure for the average user than Windows, but this advantage will evaporate if Apple doesn't take a leadership role in continuing to promote security internally, and to open itself to the security community and invite participation. A "security panel" comprised of internal and external experts, researchers, and hackers, whose goal is to make OS X as secure as possible, would be an effective way to show everyone that Apple is commited to ongoing security on the Mac.

Apple's already got a slick, public face for design and features; it needs to do the same for quality and security.

Finally, for those who are looking for something to read over the holiday weekend (those who are Mac geeks, that is), I highly recommend Amit Singh's Mac OS X Internals: A Systems Approach (also available as an e-book). This is an incredibly detailed look into Mac OS X's core architecture, and if you're interested in low-level OS design and/or Mac OS X on a technical level, this book provides 1680 pages of geek food. You won't find iPhoto tips and tricks in this book; it's all about Mach threads, IPC, design diagrams, and code samples.

Have a safe and happy Fourth of July.