At Apple's Worldwide Developer Conference (WWDC) today, Steve Jobs dropped a bombshell on the Mac community by confirming rumors that the company will transition its computers from the PowerPC architecture to Intel processors by 2007. The news was leaked in the Wall Street Journal two weeks ago and confirmed by CNet and the Wall Street Journal last week.
The reason? Power. Citing each company's processor roadmaps beyond 2006, Jobs said that the PowerPC provides 15 "units of performance" per watt, while Intel's processors will be able to offer 70 units per watt. Jobs also mentioned that they've been unable to get a PowerPC G5 processor that will run cool enough to put into a laptop, a long-standing sore point among PowerBook aficionados.
However, it's important to note that the WWDC keynote was short on hard details: no specific hardware nor specific gigahertz targets were mentioned. Support for other hardware that Apple software depends heavily on, such as AltiVec, was also not addressed. However, you won't be able to run out and buy any old Intel box and install Mac OS X, according to comments by Apple Senior Vice President Phil Schiller; Apple will restrict the operating system to Apple-sold Intel computers. It's likely that these future Macs will be able to run Windows applications better than with today's emulation software.
Jobs said that Apple has been co-developing an Intel-based version of Mac OS X for the last five years in order to keep its options open; every release of Mac OS X has been compiled in-house for Intel processors. During the WWDC keynote, Jobs demonstrated third-party applications such as Photoshop CS2 running on a 3.6 GHz Pentium 4 processor-based system under Mac OS X 10.4.1.
Apple plans to ship low-end Macs using Intel processors by this time next year, while higher-end systems for professionals will appear in 2007. Jobs specifically apologized to those who surely wished they could have a PowerBook G5 by now, so we wouldn't be surprised to find a high-end laptop high on the development priority list.
DRM in the Chip -- One aspect of this transition that could prove interesting, in all positive and negative connotations of the word, is the so-called "trusted computing" capabilities of Intel's CPUs. Little has been done with them yet, but as we understand these capabilities, they're designed to work with a Microsoft digital rights management (DRM) system. There's no telling if or how they may play into Apple's existing music or future video plans.
Making the Transition -- Developers who use Xcode should be able to make minor changes for their programs to work with Intel processors. Compiled binary applications will be able to contain the processor-dependent code for both PowerPC and Intel chips, meaning that developers can release a single program for both types of Macs. Jobs said that more than half of current Apple developers use Xcode and another 20 percent were planning to start using it soon. Not surprisingly, he suggested that everyone else get on the bandwagon, too.
Jobs also discussed Rosetta, a binary translator that turns PowerPC code into code for Intel chips on the fly. While this kind of conversion has been used for some forms of emulation by other companies in the past, Jobs indicated that Rosetta is optimized enough to avoid comparisons with the often clunky and funky operation of Classic within Mac OS X. It should be a more seamless experience for Mac users, comparable to the PowerPC transition, when the vast majority of older 680x0 applications simply ran. Jobs demonstrated Photoshop CS2, Microsoft Office, and Quicken running in unmodified PowerPC-binary form using Rosetta. Of course, just because they run doesn't guarantee that they will run well, especially for something like Photoshop, which is commonly used to benchmark processor speeds. However, it does signal to users that they don't have pay for upgrades to all of their software, as many did with the Mac OS 9 to Mac OS X transition simply to run it on a new architecture.
Apple has a long history of carrying its older users on its back as it forges across a river dividing two architectures. The change from 680x0 to PowerPC was generally good - with exceptions - and Mac OS 9 to Mac OS X was a long, slow, but ultimately successful transition as developers produced applications that could run in Mac OS 9 and Mac OS X. Even the addition of a 64-bit processor in the form of the PowerPC G5 produced relatively few problems.
Jobs also announced that Mac OS X 10.5 will be codenamed Leopard and ship in late 2006 or early 2007, around the same time as Microsoft Longhorn - just to heighten the comparison, one wagers.
Small Developer Crunch? The Intel processor transition is likely to affect smaller developers much more than larger ones. Most large software companies that create products for Mac OS X also have Windows versions. The code base can be largely identical. Smaller developers typically program for a single platform and may not have the financial or staff resources for the testing necessary.
But Apple made overtures to cater to this audience, which includes thousands of companies that currently release Mac software. Select and Premier members of Apple's Developer Connection will be able to purchase a $999 Developer Transition Kit that comprises an Intel processor-based computer and preview releases of Mac OS X and Apple software. This system won't be available to the general public, nor will it work like a normal consumer system, being geared for programming and testing. Interestingly, developers will have to return these Intel boxes by the end of 2006 - it's a loan, not a purchase.
Too Hot to Handle? In the past, Intel chips ran hotter and required more power than comparable PowerPCs. But the company has learned a lot from tuning its Pentium 4M and Pentium M for laptops, and its new dual-core architecture that has the equivalent of two processors in a single integrated circuit package doesn't double heat or power as it doubles computational performance. (Multi-core technology is apparently the near-term future of most processors, with IBM releasing a nine-core system called Cell.)
Beyond wattage figures, IBM and Intel had closed the gap on true computational measures, a previous bone of contention dubbed the "megahertz myth" when focusing on cycles per second instead of actual tasks completed. Intel has suffered a number of setbacks in the last year that have slowed their processor speed targets, but is still on track to outpace IBM dramatically in the future. IBM has had noticeable stumbles including delayed G5 deliveries last summer that pushed G5 iMacs back three months.
Gutting Sales? Technical issues aside, the real question is the reaction of consumers and professionals. Do customers respond to this announcement by embracing the current Macintosh platform more heavily, knowing there's a steady uptick ahead for processor performance with what could be a relatively seamless transition that allows them to use current software? Or will hardware sales plummet as companies and individuals decide to wait for faster machines in a year or two? (We always suggest buying what you need when you need it; there's invariably going to be something newer, better, and faster around the corner, and it's silly to wait forever until they stop innovating.)
Apple has basically conceded that PowerPC G5 chips cannot be made cool enough to be used in laptops, which means that unless Freescale Semiconductor (Motorola's spun-off chip division) can produce much faster PowerPC G4s, Apple will wind up releasing only modestly faster PowerBooks for a full two years, which could cost them quite a bit of the pro and speed-demon markets.
It's likely that Apple's roadmap shift to Intel will cause financial analysts and business writers to tell the public and institutions that Apple now is on a secure footing, no longer tied to a small fraction of a tiny part of IBM's current revenue, but is rather tying its hopes on the core business of the world's largest chipmaker. On the other hand, the stock market generally considers change to be a bad thing, and there's a distinct tinge of defeat in switching CPUs (ignoring of course, that what makes the Mac different has always been the operating system, not the technical details of the hardware underpinnings).
Even more significant is that Windows XP and Longhorn will be facing head to head challenges with Mac OS X on what is likely to be highly comparable equipment. Running a native Intel Photoshop under Mac OS X versus Windows XP will reveal more about the efficiencies of Unix and Apple's implementation than any of the apples to oranges (or Apples to Redmonds) tests yet performed.