INTEL BUYING APPLE SHOCK HORROR!

[mmcirvin]

This Robert X. Cringely column, asserting that recent developments can only mean that Intel is buying Apple in an Ahab-like plot to destroy Microsoft, is pretty stupid. I have no idea whether Intel would want to buy Apple, but I don't buy the notion that the decision to switch to Intel processors is so baffling that it can be explained no other way.

Question 1: What happened to the PowerPC's supposed performance advantage over Intel?

This is the Altivec Factor -- PowerPC's dedicated vector processor in the G4 and G5 chips that make them so fast at running applications like Adobe Photoshop and doing that vaunted H.264 video compression. Apple loved to pull Phil Schiller onstage to do side-by-side speed tests showing how much faster in real life the G4s and G5s were than their Pentium equivalents. Was that so much BS? Did Apple not really mean it? And why was the question totally ignored in this week's presentation?

Yes, Robert, it was BS. Or, rather, not an outright pack of lies, but a lot of obvious cherry-picking of stats. People do this in the computer industry; it's called "marketing". Also, that was some time ago. The G5 was actually not bad from a performance standpoint—it started out pretty competitive with fast Intel boxes—but after it was introduced, IBM showed no signs of being able to crank it up at the rate Jobs wanted, nor could they make it low-power enough for laptops. Jobs said all this fairly explicitly, if you were listening.

Question 2: What happened to Apple's 64-bit operating system? [...]

I don't know enough about the 64-bit processor market to comment on this very intelligently. I was never clear on whether the magic 64-bitness of Mac OS X actually meant anything. Maybe they can borrow some bits from the Atari Jaguar; I hear it had a lot of bits. Pass.

Question 3: Where the heck is AMD?

If Apple is willing to embrace the Intel architecture because of its performance and low power consumption, then why not go with AMD, which equals Intel's power specs, EXCEEDS Intel's performance specs AND does so at a lower price point across the board? Apple and AMD makes far more sense than Apple and Intel any day.

I don't know how accurate this is, but scuttlebutt is that AMD has potential supply-chain problems for the kinds of chips Apple wants. They'd been down that road before. If you want to do your best to avoid supply problems, best to go big.

Question 4: Why announce this chip swap a year before it will even begin for customers?

This is the biggest question of all, suggesting Steve Jobs has completely forgotten about Adam Osborne. [...] Apple's stated reason for pre-announcing the shift by a year is to allow third-party developers that amount of time to port their apps to Intel. But this makes no sense. For one thing, Apple went out of its way to show how easy the port could be with its Mathematica demonstration, so why give it a year? And companies typically make such announcements to their partners in private under NDA and get away with it. [...]

The stupidity of this boggles me, despite the kernel of truth concerning the "Osborne effect" he mentions. You've probably read about it: pre-announcing your future products too early can cannibalize sales of existing ones. Apple obviously knows this, considering how secretive they usually are about new products.

But Cringely's other claims are ludicrous; he seems to believe that everything will be as easy to recompile as Mathematica and that Apple could therefore spring this whole transition on developers with a few weeks or months to spare, and expect to have product-quality apps ready to roll in time. Developers who are not already using XCode are saying that this is going to be somewhat bumpier for them than the 68k-to-PPC transition was, and I distinctly remember that taking a while.

Cocoa/XCode apps will be almost trivial to port; other things, somewhat worse; things developed with (Freescale-owned) Metrowerks, and games that hit the system at a low level and are infested with hidden big-endian dependencies, probably quite difficult (though it might be easier to just port the Windows versions, if they exist, since these things are going to be pretty close to bog-standard PCs physically). Also, making code run fast on a different processor architecture is likely more than just a recompile. And that doesn't even take into account things like testing; ported code is new code, and we surely don't know that Mathematica for MacIntel was bug-free.

The developers do need the time, something that actually made John Gruber think that a switch to Intel could never happen in part for this very reason. And the notion that they could let every Mac software developer in the world know about this, and maybe ship them development systems, with a 12-month lead time without letting the cat out of the bag (especially given that Darwin/x86 was completely public knowledge, and MacIntel rumors had been floating around for years) is pretty stupid. We've got benchmarks on Apple's lashed-together dev box and photos of its interior already.

Companies do things like this in secret (and if my employers are reading, they can rest assured that I take NDAs very seriously), but it's partly security-through-obscurity; the huge, epochal secrets don't last long when the subject matter is something that has its own gigantic fandom and a small industry devoted to feeding the fandom rumors—a fact that drives Apple absolutely insane.

I figure that Apple knows perfectly well they're going to take an Osborne Effect hit, can't see any way around it, and announced this now because they're doing pretty well these days and can probably ride it out. Had they waited for Mac sales to slip again as a consequence of flagging processors, or for the iPod fad to fall from its crest, they'd surely die of it.

Question 5: Is this all really about Digital Rights Management? [...]

He doesn't think so. I actually agree with him here, though every other smarty-pants in the world is saying this is the secret real reason for the deal. I think that while Apple may be interested in that, it's secondary at best. Apple knows that while any DRM can and will be defeated, to make a bundle distributing entertainment DRM only has to be just good enough, and they can probably swing that with or without DRM-enabled processors.

The actual reasons for the switch are pretty obvious; nobody could give Apple the chips they needed in the quantities they wanted. The only reason it's a surprise is the sticky question of whether these problems outweigh the real risks of the switch. I certainly don't know. Apple evidently thinks so.

Having uncovered all these supposedly baffling anomalies, Cringely announces that the only good explanation is that Intel is buying Apple so that they can wrest the PC market away from Microsoft, which insulted their sacred honor, is now their sworn archenemy and must be destroyed just like the six-fingered man who killed Intel's father. Meanwhile, people accuse Steve Jobs of making decisions based on pique.

Comments

[mmcirvin.livejournal.com]

Developers who are not already using XCode are saying that this is going to be somewhat bumpier for them than the 68k-to-PPC transition was, and I distinctly remember that taking a while.

I should add, for people who might be freaked out by this, that in one way the transition is likely to be smoother for users than the 68k-to-PPC transition was. One thing that slowed down the early PowerPC Macs running 7.x was that large chunks of the operating system were running under emulation as 68k code. The frequently-used stuff had all transitioned by OS 8.0, and the fraction of 68k code in the system gradually declined as time went on (but I don't think it ever dwindled entirely to zero until Mac OS X; these transitions always seem to be brought to a full close only by the next big transition).

That shouldn't be the case this time around, since Apple's been compiling Mac OS X as native Intel code from the beginning.

[ckd]

On 64-bitness, the big win of that is getting processes that can use more than 2GB of address space. In some areas (*cough*bioinformatics*cough*) that's a real need, which is why the Alpha was the processor of choice for those applications before Digicompackard killed it off.

The Itanic-AMD64 fight hasn't necessarily helped any.

[graydon.livejournal.com]

I have a hard time considering this entire situation to be anything other than proof of mac people having some amiga-like cult elitism surrounding their hardware. it's a chip. the other chip is also a chip. they fetch, decode, execute, and retire instructions. the intel one will do the same thing.

besides, there's no need to worry about a year worth of "osborne effect" when the music side of the business (which doesn't give a hoot) is growing at 300%.

[ammenemes3.livejournal.com]

I think Apple just wanted to give IBM one last good solid kick in the teeth, for good measure.

[mmcirvin.livejournal.com]

Hannibal of Ars Technica (http://arstechnica.com/columns/mac/mac-20050608.ars) says everything I was trying to say, only more knowledgeably and elegantly.

He also muses on why this shook the Mac fans so badly. I do think he underestimates the pull of Mac OS X as opposed to imagined hardware differences. Personally, I was souring on Macs and pondering decamping to something like an Intel Linux box just when OS X started to get interesting. I know a lot of old Mac enthusiasts never warmed to OS X and insist to this day that it violated everything the platform stood for, and some points they made concerning the UI and file system were valid (especially in the early versions), but for me it was fascinating to see it grow, and the idea of a fancy desktop OS that was also a Unix-like OS was extremely attractive to me, since I was pretty familiar with Unix.