[Pc_Support] Mactel? Due to product focus? Volume? Yield issues? Other?

[Bryan J. Smith <b.j.smith at ieee.org>]

Well, it's official, and it kinda shocks even me.  I figured Apple might only be toying with the idea, or looking a trial.  eWeek's confirmation:  
http://www.eweek.com/article2/0,1759,1824695,00.asp  

But given MacOS X was built for both x86 and PowerPC, and even released public versions through Developer Release 3 (DR3) long ago, it was probably inevitable.  I'm just still stratching my head because I thought the IBM PowerPC 970 (G5) was the ultimate balance of efficiency, power/thermal, performance and 64-bit transition.

But in reality, I have to assume some serious factors brought Jobs & co. to this point.

First off is the product focus.
IBM is going to dedicate its fabs to the volume, and that's now PS3 and Xbox 360.
I have to assume issue #1 is probably the sheer devalue of Apple's importance.
And possibly the entire purpose for the PowerPC variant of the Power-line.

Secondly, it could be yield issues, possibly low-power version?
The whole dream of the PowerPC 970 (G5) was supposed to be a sub-20W variant
at 66% the leading edge desktop version's performance.
But that promise is now years old, and both Intel has brought back the Pentium III core in a 21W version at over 2GHz known as Pentium M, while AMD's latest 1.8-2.2GHz (Athlon 64 3000-3500+) Winchester 90nm processors are even faster for only 10W more.

Third, it could be other issues like maybe the 64-bit transition.
Maybe Jobs was seeing not only a conversion issue by sticking with PowerPC 970 and the Power-lineage to 64-bit, but that there might be a kernelf maturity issue as well.
PowerPC 970 was supposed to be a more simplified split 32/64-bit transitional approach than 48-bit/PAE52 "Long Mode" of x86-64 that had to deal with segmentation and compatibility of the translation lookaside buffer (TLB) of i486 on-ward.
But now it seems the more likely case is that 64-bit is now well understood on x86-64 (AMD64/EMT64), and Jobs saw its adoption down the line versus what IBM planned to offer in volume, and decided to just kill two transition birds with one stone.

Lastly, it could be sheer software optimization and other factors.
MacOS X relies on GCC, and that means the optimizations of the even the 32-bit PowerPC target (let alone support of Altivec extensions) continue to lag well behind x86/x86-64.
This has been well documented, and it does not seem to be improving anytime soon.

All-in-all, the switch to Intel is a bit of a shock, but not without careful consideration.
It's hard to beat the economies of scale of x86/x86-64, and IBM's fabrication and volume focus is probably leaving little time to address Apples more limited volume needs.
The lack of a portable G5 is probably a sign of this reality, and the 64-bit and GCC issues problem are icing on the cake.

Which begs the question, will I be able to run MacOS X on any x86/x86-64 PC?
Or will MacOS X only run on x86/x86-64 PCs with a specific hardware/firmware option?
Or even better yet, will Apple _finally_ smack the the PC BIOS/firmware as well as Microsoft silly with a _real_, _intelligent_, _independent_ firmware/pre-OS configuration management solution and not stupid, OS version-limited/software-driven firmware ?
Hmmm, maybe Intel sees this as a way to finally address that major issue that Microsoft and the PC BIOS/firmware vendors still haven't.

And what about AMD systms?
If anything, the seemingly Intel-only partnership really and strongly tends to suggest this is a volume-based decision where IBM just isn't going to care.

In any case, if it either runs on generic systems (especially AMD), or Apple builds a cool PC for a "few bucks more," I might find myself buying one for a triple boot.
And I know I'll probably be running MacOS X as my primary desktop, with Windows for that rare application (if it doesn't run under VirtualPC, which might be free), and the occassional Linux software (if it also doesn't run under VirtualPC).


--
Bryan J. Smith   mailto:b.j.smith at ieee.org