Backwards compatibility is the reason why PCs were still running DOS-based versions of Windows less than a decade ago (Windows 9x). Before everybody was concerned with security (outside of enterprise users) we were always hearing about stability. OS9 wasn't very stable either because of it's ancient roots. It is important to some, but not important to most. Apple has handled all their transitions (68k -> PPC, OS9 -> OSX, PPC -> Intel, etc) quite well and maintained backwards compatibility for at least five years in nearly all cases. That's plenty of time. If the software relied upon this, blame the people who didn't update the software (and also keep an old system around).
Amen to that.
If for your mission critical tasks/data you rely on a piece of software written by some unknown company, released on a closed source basis, and don't have enough money to persuade them to port it to a different/newer system, given that they are still around... well, you are not being a wise IT manager.
If a different CPU is a worry to you in terms of backward compatibility, you are simply not doing it right.
Frequent architecture changes are GODSEND to ensure that the last managers who believe in the cornflakes approach to software either change their mind or go doing something else.
Speaking of ancient roots, the x86 architecture is holding us back too. The SIMD extensions sets and x86-64 extension set along with the extra cores and SMT are what make it so powerful. Synthetic benchmark performance improvements from one architecture generation to the next go unnoticed by the majority of people too because they never utilize that extra power. Speaking of power, I still believe that PPC is a better architecture than x86 but that's just my opinion.
Yeah.
Point is, we must differentiate "architecture" and "products".
x86 is no better than PPC or ARM.
Only, Intel delivers better products by a mile, because they have the resources and the technology to get a good product out of an average architecture.
PPC: great architecture, but is there a chipmaker in the world that makes PPC CPUs that are not stuck in the pleistocene when it comes to manufacturing process, die size, extensions, etc?
But, here we go: we've had HUNDREDS of different ARM-based chips, as you know ARM is slightly different from other architectures because it works on a license-only basis.
Does Apple have the resources to make some great ARM-based chips?
Right now, not really.
But Apple has a history of buying entire factories and, perhaps - not sure -foundries.
Apple has lots of money stashed away.
Is it possible that Apple can make a kickass general purpouse ARM CPU in the future?
HELL YES!
I also believe that the Intel transition was a great move on Apple's part.
+1
As far as switching to ARM in the future and having a unified single OS. Well as already mentioned, iOS already is OSX. If you've jailbroken your device and used the terminal, the file system is very similar (it is BSD unix afterall).
I don't think that the article is saying that all the bloat from the desktop release would carry over to the phone. I've always thought that Apple's intent was to make OSX more iOS-like and unify the two in the future. Even MS's Windows 8 is going to have an ARM release.
On the PC side, Intel has the advantage today, but from a performance-per-watt standpoint ARM kills it on the mobile side. A while back I saw a test where a single core ARM CPU was nearly equal to a single core Atom CPU except that it consumed a fraction of the power. Computing will move on from x86, it has to. Mobile is huge and x86 doesn't get along with mobile. Having a single OS and single computing architecture for both mobile and PC just makes a lot more sense.
I think you are very much right, Sir.