Daniel Eran Dilger
Random header image... Refresh for more!

WWDC 2008: Is Mac OS X 10.6 the Death of PowerPC?

Daniel Eran Dilger
Ever since the Intel Mac transition was announced at WWDC 2005, pundits have been seeding worries that Apple would immediately pull the plug on PowerPC Macs. In 2006, the rumor mill stated Leopard would be Intel only; will OS X 10.6 finally mark the end of PowerPC support?

Leopard on PowerPC.
Two years ago, I presented why Apple wouldn’t release Leopard for Intel Macs only. That was in response to speculation by Mark Stephens, writing under the name Robert X Cringely, which insisted that Leopard’s “new OS X Kernel won’t be backward compatible to PowerMacs” because if Leopard made those machines run faster and better, their owners wouldn’t have any reason to buy the new Intel Macs.

Incidentally, Stephens also wrote that Apple would devote its scarce resources to work on a “Red Box” for native Windows compatibility while moving Mac OS X to the Linux kernel to speed up its “integer math.”

The main evidence supporting my confidence of a Universal Binary of Leopard was that excluding PowerPC support would cut away the vast majority of its potential market. At the time, it appeared that Apple would release Leopard in early 2007. I pointed out that by that time, Apple would have only sold around 3 million new Intel Macs, but have an installed Universal base of 14 million Macs less than four years old, and over 20 million Macs that were able to run Mac OS X.

Sure enough, Leopard did ship for both PowerPC and Intel Macs, and Apple earned $170 million on strong sales in the first quarter of its release. The company estimated in January that 19% of the installed user base had upgraded to Leopard, indicating significant sales to PowerPC users.

Unraveling The PowerPC Obsolescence Myth

More Intel Macs Sold Than Anyone Anticipated.
At the same time, Leopard didn’t make it to market until October 2007. During that delay, Apple’s new Mac sales continued to grow rapidly. By the time Leopard went on sale, the company had actually sold closer to ten million Intel Macs. Since Leopard’s arrival, Apple has sold another 6.5 million Intel Macs in just nine months.

That means Apple now has an installed base of around 16 million Intel Mac users, but only about 7.5 million PowerPC Macs that are five years old or less. I had earlier calculated that Apple wouldn’t likely accumulate more Intel Macs in its installed base than PowerPC Macs until 2009. I didn’t realize that new Mac sales were going to double within three years. That means the main evidence supporting the likelihood of future Universal Binary releases of Mac OS X is waning.

Increasingly fewer PowerPC machines are less than three or four years old, which makes them less likely for consumers to want to upgrade beyond Leopard, particularly since Leopard still has years of supported, useful life left in it.

Drawing the Line at Four Year Old Hardware.
Last fall, I detailed Apple’s historical line of support drawn for older machines back through Mac OS release into the Copland days, when Spindler’s Apple tasked its engineers with the unreasonable expectation of developing system software that could run on decade old hardware. Ten years ago in 1997, it seemed like a bold step for Apple to release Mac OS 7.6 with support limited to “32-bit clean” Macs. That still allowed it to run on the then eight year old Mac IIci from 1989.

At the same time, the New Apple developed a modernized Mac architecture around the PowerPC G3 in 1997, and made the new minimum supported machine for the upcoming Mac OS X, released four years later in 2001. Progressive versions of Mac OS X pushed the line ahead slightly, with 10.3 requiring Macs with New World firmware, and 10.4 dropping some models lacking Firewire support. Leopard 10.5 officially drew the line a 867 MHz processors, which effectively retained a four year window of support for existing hardware.

For 10.6 to be released in early 2009 as Intel only, Apple would be cutting support for all Macs sold before 2006, along with some PowerPC models sold in that year. If 10.6 were a conventional release, such a short window of backward support would make less sense. However, the Ars rumors of Snow Leopard, confirmed by independent reports I have received from inside sources, suggest that Mac OS X 10.6 will not be a typical release.

Leopard and the History and Future of Mac OS X on PowerPC

Leopard and the History and Future of Mac OS X on PowerPC

Not Another Feature Release.
I like to point out that consumers don’t like to buy software, and certainly not with the enthusiasm they buy hardware. If Apple, Microsoft, or any developer released a new version that was primarily a bug fix, consumers would complain that they should be getting it for free, because the original version was broken. For this reason, software vendors feel compelled to add new features that can be touted in marketing, rather than just refining and improving the existing ones. Somewhat ironically, consumers then complain about the plague of featuritus and outstanding bugs not being addressed.

Every major reference release of Mac OS X has been given big new features with splashy marketing names: Exposé, Dashboard, Time Machine, Spaces, and so on to convince users of the need to buy the upgrade. There’s plenty of new applications Apple could add to Mac OS X, but there are increasingly fewer it needs to add. Most ideas for new apps could be made part of iLife, iWork, or sold independently.

That lends credence to the idea that 10.6 will break the mold of previous releases to become an unusual refinement release, one that takes special advantage of recent hardware to advance performance and simplify the system. Also supporting this idea is the reality that Apple’s active installed base is now mostly Intel based, thanks to the rapid increases in Mac sales since 2006.

Independent sources have also confirmed the internal name of Snow Leopard. Whether Apple uses that name in marketing is yet to be determined, but it does indicate that the next major revision of Mac OS X will accompany Leopard rather than replacing it outright, creating a scenario where PowerPC machines continue running Leopard while Intel Macs begin running a specialized Snow Leopard version optimized for their hardware.

There won’t be any reason to complain because the Intel only version won’t be advancing big new marketing features. Further, PowerPC code will continue to run on 10.6 using Rosetta.

So what will 10.6 advance: the end of Carbon? The next article will take a look: WWDC 2008: Is Mac OS X 10.6 the Death of Carbon?

WWDC 2008: New in Mac OS X Snow Leopard

WWDC 2008: Predictions & What to Expect: Mac OS X 10.6
WWDC 2008: Future UI Designs in Mac OS X 10.6
WWDC 2008: Moscone West Spy Shots!

I really like to hear from readers. Comment in the Forum or email me with your ideas.

Like reading RoughlyDrafted? Share articles with your friends, link from your blog, and subscribe to my podcast! Submit to Reddit or Slashdot, or consider making a small donation supporting this site. Thanks!

Technorati Tags: , , , ,

  • Galley

    If it does not add any significant new features, then it may indeed be a free upgrade, just like 10.1. If it does cost, let’s hope it’s no more than $29.

  • anonymous500r

    I really hope that they resurrect the POWER architecture – especially now that IBM’s brought out it’s POWER6 architecture (4.7GHz – mmmm.) If Mac OSX is as portable as it’s claimed – then why not have a multi-architecture approach? ARM in the MacBook air-type machines, Intel in the MacBook’s and Minis and POWER6 in the Mac Pro and Xserve? Now that would be a good WWDC.

  • solipsism

    @ DED,

    Could this *new* version be ONLY for new Macs as are Intel, 64-bit and protected with HW authentication. Apple knows that it can’t prevent the OSxProject from hacking their code or Shystars making money on Mac clones by sueing them, but they can out tech them. Right now, they can’t compete, but HW is getting faster and cheaper, someone may come along who can make a convincing argument to by an unofficial Mac clone, especially in countries that Apple has progressed slowly.

    Perhaps like Icarus, Apple moving to Intel was an escape from the dead-end maze of PowerPC, but by not taking further precautions the ubiquitous x86 could eventually hurt Apple’s Mac line. If they were to start making every new line with optimized code that was faster, marginally more stable, smaller and protected to these new machines then they would have a leg up on the competition and the thieves.

    From a marketing standpoint, this would encourage current Mac owners to upgrade sooner to get the better optimized code. It would also assist with the potential future issue of widespread hacking and illegal use of Mac OS X on non-Apple HW. The MacBook, MacBook Pro and Mac Pro lines are all up for a case revision. What better way to start it.

    PS: If they continue offering Intel-only for another OS X cycle to get past the 4 year obsoletion rule, I’d wager the next UB version will be called Lion and the x86 version will be Mountain Lion.

    @ anonymous500r,

    While Mac OS X is highly portable, Apple would still have to support it for 3 different architectures. That is a lot of work, a lot of debugging and a lot of extra code. If Apple can support only x86 for the Mac OS X then things become a lot easier for them and much better for us as the code is more stable, more refined and easier to produce.

  • lmasanti

    “While Mac OS X is highly portable, Apple would still have to support it for 3 different architectures.”

    Maybe this is the “substance” of the rumor: Apple just wants to support 2 architectures.
    Until last June, they were PowerPC and Intel.
    From Junes, PowerPC, Intel and ARM.
    Soon they could be only Intel and ARM.

    But I do think that 10.6 will still support PowerPC.

  • Pingback: WWDC 2008: Is Mac OS X 10.6 the Death of Carbon? — RoughlyDrafted Magazine()

  • http://coderad.net StrictNon-Conformist


    Portability isn’t the big factor.

    Neither is the relatively expensive proposition of supporting another CPU architecture (the POWER architecture isn’t THAT different from currently used PPC Macs; it’d just be local motherboard issues, firmware issues, and recompiling for the new processor using a compiler flag option for optimization only).

    No, the biggest issue is the business reality that IBM’s most recent POWER systems, however fast they may be, don’t have the sales numbers to make it cost-effective for Apple to be bothered: it just doesn’t fit in with Apple’s current overall strategy of supporting and selling products that have sufficient mass-market appeal that allows them to ship commodity hardware with nice software that’s the most important value-added part of the equation.

  • http://opinion-nation.blogspot.com/ philipmach

    I should point out that portability of the OS code is the reason that Apple was able to transition to Intel so fast. The same is not necessarily true for apps; maintaining a PPC version encourages developers to keep maitaining PPC builds as well. Apple will probably keep maintaining a PPC version for as long as it takes to keep customers happy for no better reason than to keep Intel on notice that they could switch back.

  • http://opinion-nation.blogspot.com/ philipmach

    As for the Power CPUs, examine the cost and power specs before you get too excited. IBM’s mass-market parts are these days aimed at the embedded space and games boxes, and a personal computer is somewhere between the two sets of requirements: not quite as price sensitive as embedded / toy; a lot more price sensitive than big iron servers.

  • PerGrenerfors

    I expect that 10.6 will drop the G4 completely but keep the G5. Lots of Power Mac G5 and later iMac G5 are still okay machines. In addition, the hardware requirements will be really straightforward: G5 or Intel processor, minimum 512 MB RAM.

  • Pingback: Nanchatte Technojunkie » Blog Archive » W-3 : Apple Rumour Mill - The Hamsters are hard at work()

  • gus2000

    I envision a compromise between lugging around the anchor of PowerPC and dropping it completely: instead of making one OS build using all Universal Binaries, make two separate builds (Intel and PPC) instead.

    There’s great benefit in the current one-DVD-fits-all scheme, but PPC will be dropped eventually and the split seems like a logical intermediate step. It would allow the Intel code base to move forward unencumbered by legacy, and signal to the PPC owners that the End Is Near. Obviously some features would be omitted for PPC, but at least the G4/G5 owners would be able to have a current, supported release through their end of useful life.

    I’m still using a G4, which is quite usable but showing it’s “Moore’s Law” age. If I really want to continue to have it useful for another year or two, OS X should trade future features for a PPC-optimized release that can maximize performance with limited system resources.

  • http://johnsessays.blogspot.com John Muir

    Have a look at these reported screenshots of the Snow Leopard WWDC seed:

    This one in particular:

    That Application Type column looks reassuring!

  • gus2000



    It looks like the current builds of 10.6 are Intel-only.

  • tehawesome

    Is there any huge difference from a developer’s point of view in terms of writing an app on an Intel-only version of Snow Leopard, then compiling and distributing a PPC version? My guess is that Grand Central and OpenCL are going to be Intel-only, but, as with the current Core APIs, will degrade cleanly so that without the appropriate hardware, you just don’t get the new features. As it stands, it’s important to qualify that the current 10.6 build is Intel-only (and we’ve seen Apple being sneaky before about maintaining parallel builds for different architectures: can you say Marklar?) and not rule out a PPC version, though it does rather seem that this drives another nail into the coffin. Still, I’m writing this on a G4 Cube running 10.4, and it’s increasingly looking like I’m not going to ever run 10.5: a Nehalem Mac Pro running 10.6, however…

  • Pingback: Mac OS X 10.6 "Snow Leopard" to arrive Jan '09 - Page 6 - MacNN Forums()

  • Pingback: Myths of Snow Leopard 1: PowerPC Support — RoughlyDrafted Magazine()

  • Pingback: Myths of Snow Leopard 2: 32-bit Support — RoughlyDrafted Magazine()

  • Pingback: W-2 : Apple Rumour Mill – The Hamsters are hard at work « The Nanchatte Technojunkie()