According to Hoyle...
Mac OS X 10.6
Snow Leopard: The End of the Line for the PowerPC
August 2009
by Jonathan Hoyle
jonhoyle@mac.com
macCompanion
http://www.jonhoyle.com
Next month, Apple
will be releasing its next generation operating system: Mac OS X 10.6 Snow
Leopard, a replacement for its previous Mac OS X 10.5 Leopard. I had intended this month to review
10.6 (within the confines of Apple's NDA), but found it essentially a repeat of
last year's article on Snow Leopard following WWDC '08 Rather than reiterate the same things
again, we will wait until its official release, whence I may devote a column to
review it fully. This month, I'd
like to talk a little bit about the PowerPC processor, as its reign dies with
10.6.
PowerPC:
1993-2009
One
might quibble with the dates here. The AIM (Apple-IBM-Motorola) alliance created the PowerPC in 1991, and
the first Macintosh computer using this processor (the Power Macintosh) was sold in 1994. Although Apple ceased production of PowerPC-based
Macs in 2006, PowerPC chip itself is still being used today in other products. So why do I use "1993-2009" as the start and end dates in the above paragraph header?
I
chose 1993 as the start date, since that is when the first PowerPC development
tools became available for the Macintosh consumer. (I refer more specifically to the beta release of Metrowerks' offering, not
that God-awful cross-compiling strategy that Apple was using at the time.) For the end date, I picked Snow Leopard's release (this year), rather than when
Apple stopped producing Power Macs. I did this because the PowerPC is still a significant percentage of the installed
base, and the chip is still supported in as far as the fact that the current OS (Mac OS X 10.5 Leopard) still supports the
PowerPC G4 and G5 processors. That will end with the arrival of 10.6.
So this is it for
the PowerPC processor. This is not
simply a Snow Leopard statement, but an Apple statement in general. Apple's recently released Final Cut Pro
7 and Logic Studio are Intel-only, despite its running on 10.5 Leopard (a
PowerPC-supported operating system).
This
should not be terribly surprising news. After all, it was three years ago this summer that Apple officially
killed off the PowerPC line of processors. And although there are a fair number of G4's and G5's still
out there in the field, they are a diminishing breed.
Truly Seamless
This is not Apple's first processor transition. Apple's
transition from 68K to PowerPC was an amazing success. If you think Apple's current transition
from PowerPC to Intel was seamless, then you need to go back in time by a
decade to see was seamless really was. The 68K emulator on the Power Macintoshes was fast and extremely versatile. Although it did not support the 6888x
Math co-processor, most anything that ran on a contemporary 68K machine (such
as the Centris 610) would also run on a Power Mac.
And
not just applications. Plugins,
dynamic libraries, code resources and other programming components can be 68K
and run inside a PowerPC-based application. This is actually one of the significant differences between the
68K/PPC transition and the PPC/Intel transition: plugins and libraries must
match precisely the architecture on its hosting OS X app: 32-bit PowerPC,
64-bit PowerPC, 32-bit Intel or 64-bit Intel.
Another
difference is performance. From
nearly the beginning, the 68K emulator's speed was quite reasonable. By the end of the 90's, the 68K
emulator on a Power Macintosh was running code faster than that code would on any previously sold 68K Mac.
That's not quite true with PowerPC->Intel. For example, Rosetta emulator on Intel is not
too bad, but a high end G5 still smokes it, even three year later.
Users
on PowerMacs in the days of the first transition, often did not know if their favorite app was running
native PowerPC or was 68K-emulated ... frankly they didn't care. Why should they?
The Real Hero
Yes,
the 68K to PowerPC transition was an amazing success for Apple ... perhaps more successful
than it had any right to be. Apple
had been very slow with its development tools, and with no native PowerPC
applications, there would be no compelling reason for Mac users to leave their
68K architecture. The dominant Mac tool providers at the time was Symantec, makers
of Think Pascal and Think C,
but surprisingly, they had no interest in investing in the new PowerPC processor. The only other
option was Apple itself, but its development tools
were too high end for the casual developer. That might have been the end of it until ...
In
walks Metrowerks, with a product that would eventually be given the name
CodeWarrior. This development environment was
simple to use, allowed users the flexibility to select their choice of front end programming language
(Pascal, C or C++), as well as selecting their choice of back end (68K, PowerPC or both). More than any one single force,
Metrowerks saved Apple's bacon and paved the way toward a successful PowerPC
platform. In the late 1990's and
early 2000's, Metrowerks CodeWarrior became the dominant development environment,
eclipsing Symantec's Think tools and Apple's MPW. It's
hard to imagine what the Mac world would look like today had the PowerPC failed.
Thus, it's no overstatement to say that the
success of the PowerPC platform on the Mac, was strongly thanks to the birth of
CodeWarrior. Ironically, the death
of the PowerPC was also due to the death of CodeWarrior. Despite the protests of its user base,
Metrowerks sold off its Intel compilation tools in 2005, just weeks before
Apple announced its Intel transition, turning CodeWarrior from a monopoly to
irrelevant over night.
PowerPC Evolution
The
first Power Macintoshes (6100, 7100, 8100) came with System 7.1.2. System 7 evolved into System 7.5, then
Mac OS 8 and 8.1. From 1994
through 1998, Apple supported its operating systems on both its 68K and PPC
platforms. Then in late 1998, Apple
introduced Mac OS 8.5, which dropped support for 68K Macs. In 1999, Apple introduced Mac OS 9, its
final Classic offering, as work began in earnest on Mac OS X. With Mac OS X (public beta in 2000,
released in 2001), Apple ceased support for all pre-Steve Jobs PowerPC
processors, requiring the G3 or later processors.
All
this time, the PowerPC processor continued to hum along. The initial PowerPC processor, the
601, ran at 60 MHz, but before too long,
speeds continued to grow ever faster, with the PowerPC G4 chip
running at 1.5 GHz and faster. All these chips were created by the AIM consortium. But despite
speed and chip improvements, however, these AIM processors were not improving with the same velocity as
the chips created by their rival Intel equivalents. Apple had
always prided themselves in being faster than their Windows rivals, but found
themselves falling behind. Apple felt that Motorola was not holding up its end of the AIM bargain.
At
the 2003 WWDC, Steve Jobs announced that the Macintosh line will move to a new
processor, the PowerPC G5, made not by AIM, but exclusively
by IBM. The expectation was that IBM would be
able to keep up with Intel, whereas AIM hadn't. As it turned out, Apple would be disappointed again,
with IBM failing to deliver on its promises as
well. Specifically, Steve Jobs
had promised (at the 2003 WWDC) that they would deliver a 3 GHz G5 system by the following
year. Unfortunately, Jobs had to eat his words at the 2004 WWDC, and was
uncharacteristically sheepish about publicly failing to meet this
commitment. Steve did not want to be in this position again. There would be a fast PowerPC chip in 2005
... "or else".
"Or Else" happens
I can only imagine what the back room discussions at IBM were
like in 2004: Yeah sure, Steve Jobs can threaten all he wants, but can he do about it?
He just had his Macintosh line make the big transition from the AIM processors to our G5. What was he going
to do now? What choice did he have
but wait? What's the alternative? It's not like he's going to switch to Intel.
That would be madness, right? I'm sure there were plenty of snickers then.
Say
what you want about Steve Jobs, but don't piss the guy off. And definitely don't make him look bad
in front of his own developers at WWDC. IBM called Steve Jobs' "bluff", and paid the price for
it. IBM invested $3 Billion in this G5
venture, with Apple being its primary customer. When IBM broke its commitment to keep up, and it ended up
with a very expensive lesson learned.
The
rest, as they say, is history. Apple announces
its transition to Intel at the 2005 WWDC. At the following year's WWDC, the final PowerPC-based Macs are
discontinued. In only a single year, and Apple was out of the
PowerPC hardware business. But because the installed base in 2006 was still heavily PowerPC,
Mac OS X 10.5 Leopard (highlighted at the 2006 and 2007 WWDC's)
needed to continue to support both PowerPC and Intel-based Macs. However, that is no longer the case
today. When Snow Leopard is released, it will
have been more than three years since the discontinuation of PowerPC-based
hardware. Three years is a long
time. It seemed pointless to Apple
to continue to support these old machines in a new OS.
10.6: Not a
Big Loss to PowerPC Users
In
fairness, 10.6 wouldn't have been of large interest to PowerPC owners
anyway. The reason for this is
that many of the existing Power Macintoshes are still running Mac OS X 10.4 Tiger, and
so haven't even moved up to 10.5. 10.5 Leopard runs a bit sluggishly on all
but the fastest of G5's, so making the OS upgrade was not so desirable. Furthermore,
moving to 10.5 means no more Classic. If you had no need
for Classic, there was little reason to hang onto your Power Macintosh, and you
probably had already transitioned to an Intel Mac over the past four years. But, if you still
hadn't kicked the Classic habit, you had no choice but to hang onto your PowerMac/10.4
system. 10.6 would not have changed that, even had it supported the G5.
For
those who have already made the jump to Intel-based Macs though, 10.6 Snow Leopard is
a much more compelling proposition. If you are on 10.5 Leopard (as most Intel-Mac users are), your $29
upgrade fee buys you performance improvements and OS enhancements that will
make your computing experience much smoother. If you are still on 10.4 Tiger, then your costs are no more
than the $129 that you would have paid for Leopard, but instead you get Snow
Leopard (essentially a better upgrade for the same price).
One
of 10.6's advantages that it has over 10.5 is the ability to optimize on a
single hardware family. On 10.5,
all its code had to allow for the possibility of running on either PowerPC or
Intel; for 10.6, it could rely exclusively on Intel processors. This makes it easier for the developer
targeting 10.6, since no Macs made prior to 2006 would need to be tested.
Conclusion
Soon, references to the PowerPC processor will
seem as nostalgic (and as irrelevant) as do references to the 68K family of
processors. As I write this, I
still have a PowerBook G4 by my side (shut down, not been open for nearly a
week), as well as an old Power Mac G4 upstairs (that I use as exclusively as a server). I
have mostly weened myself off of old
Classic applications, and find little reason to go back. Still, there is a nostalgic side of me
that keeps me from getting rid of my G4's entirely, even if such delays causes me to
lose whatever profit I might glean from selling now
.
On my Intel-based Macs, I am looking ever
forward, excited to run prerelease versions of Snow Leopard and take advantage
of the better performance. I am
ready to say Hello to the future ... while admittedly still hesitant to saying Goodbye to the
past ... despite knowing it is only a matter of time.
As someone once told me: All things are only a
matter of time.