Will, you're pretty drastically oversimplifying the situation. A computer is more than just a CPU. Apple's designs are entirely home-built, everything from the SATA controller to the PCI-X bridge. There's no way Mac OS X for IA-32 will be able to boot on anything other than a Mac. I just can't imagine how that would be possible.

As an Intel processor designer, I must say I am happy to see it. Even though I dabble as a computer artist, I never developed the typical preference for Macs.

Jeff, you may well be right, but looking at the way the Linux cult whips out homebrew device drivers, and I can definitely see a hacked OS X running on non-Apple stuff in the future. If they can hack an XBox or a Tivo, there's no reason why they can't figure out how to reverse-engineer a Mac.

The two examples you cited (XBox and Tivo) are examples of hacking hardware to run different software. I suspect it will be more difficult than you think to do the reverse, at least in any consistently usable form.

I wouldn't be surpised if some enterprising soul is able to get some subset of OS X running on a white-box Intel system. Nor would I be surprised if some group of people are able to mantain a half-assed working implemention. I would be surprised if the numbers ever came close to anything that remotely endangered Apple's hardware sales, simply because the vast majority of people who want Macs would not put up with the hassle.

IMHO: Stoked by the outrageous success of his non-computer products, Jobs is taking one last stab at his first longtime failed obsession. After protracted, costly attempts to seize the small PC market -- the cube, the flat screen, the I-Mac, G-Whatever -- Jobs has finally come to grips with the fact that the computer-buying public is now too well-informed to buy a machine that's strong on style and intuitive design and weak every other way: too high cost, too few games can be played on it, too costly to upgrade and, in many cases, with built-in obsolescence. His solution will ultimately be to offer a machine with all the higher-cost interface benefits of Mac, but which can be loaded with either Mac OS software, or a Windows OS, not using an emulator but as a native OS. The current switch to Intel hardware is just part of this process.

Not counting on Jobs to get this right. Wasn't one of his first actions when he came back the cancellation of clone licenses to UMAX and Motorola?

"Apple's designs are entirely home-built, everything from the SATA controller to the PCI-X bridge."

That's currently, because there is no other PC that runs on their CPU. Once they switch to Intel don't you think they'll try to save some money by using standard (or near standard) hardware?

Sure, you won't be able to install OSX on some 200 dollar Wal-Mart special, but I bet you'll be able to save a few hundred dollars by buying all the same pieces that Apple uses and putting it together yourself.

Will, I've got no doubt that the Linux cult — perfectly apt description of that freak show, by the way — will have Linux running on the new Macs within days of their release next summer. And I'm sure they'll have Darwin running on their grey-market, fell-off-a-truck, home-built PCs, if they don't already.

But Mac OS X is a wholly different thing.

And Matt, I don't know what the developer NDA covers, so I won't go into too much detail on this, but I have a source who has provided me with some details on the IA-32 machines that are going to start shipping to ISVs in a couple of weeks. They're Power Mac G5s with almost totally stock system boards and new, air-cooled IA-32 PMUs. The U3H memory controller and bridge ASIC has been altered to match the bus timing of the IA-32 processor, but that's all. Everything else on the system board is exactly the same. The internal components are all still connected via Hyper Transport through the K2 ASIC and the PCI-X bridge chip. The PMUs have 3.6 GHz Pentium 4 processors on them, but these will definitely not be the processors that Apple ships next year. The processors will be IA-32-instruction-set-compatible, but they will not be Pentium chips. They're going to be specially designed processors that Intel delivers to Apple but to no other customers, binary compatible with the Pentium family but not identical to any off-the-shelf microprocessor. For lack of a better name, I've taken to calling them "G6," but that's totally my own invention and not meant to be in any way authentic. It's just my own shorthand.

All that information comes from a source that I trust, and one that I'll protect by not saying anything more about him. Hell, I won't even swear to you that it's a him.

Bottom line: Just as Apple has been planning for half a decade to make Mac OS X microprocessor-agnostic, the design of the Power Mac G5 system was based around the idea of isolating the processors from the other components on their own bus, making it possible to swap out CPUs with only minimal changes to the other chips on the board. In this case, only one chip has to be changed, and those changes are slight.

All indications at this point are that Apple has no intention of changing its business model one iota. They're simply going to a different microprocessor vendor. In that way, this transition is no more drastic than the switch from Motorola to IBM when they went from the G4 to the G5. And it's only slightly more drastic than that to the developers. The whole porting guide is only about a hundred pages long, and it's crammed full of detailed examples.

If you can't beat 'em, join 'em.
Intel's process technology can't be beat. AMD, IBM, Freescale (formerly Motorola) and everybody else are constantly trying to catch up. Running with Intel processors will boost Mac speeds and create a lot of potential spillover benefits. The biggest problem for Apple is not the tedhnological challenge; it is the financial challenge of surviving a year or two with extremely sluggish PowerPC Mac sales.

Facinating stuff. Thanks, Jeff.

I just looked through the porting guide. Appendix A talks about Rosetta. Here's the hard facts, directly copied and pasted:

Rosetta is a translation process that runs a PowerPC binary on an Macintosh using an Intel microprocessor—it allows applications to run as nonnative binaries. Many, but not all, applications can run translated. Applications that run translated will never run as fast as they run as a native binary because the translation process itself incurs a processing cost. How compatible your application is with Rosetta depends on the type of application it is. Applications that have a lot of user interaction and low computational needs, such as a word processor, are quite compatible. Those that have a moderate amount of user interaction and some high computational needs or that use OpenGL are, in most cases, also quite compatible. Those that have intense computing needs aren’t compatible. This includes applications that need to repeatedly compute fast Fourier transforms (FFTs), that compute complex models for 3-D modelling, or compute ray tracing.

Rosetta is designed to translate currently shipping applications that run on a PowerPC with a G3 processor and that are built for Mac OS X.

Rosetta does not run the following:

* Applications built for Mac OS 8 or 9
* Code written specifically for AltiVec
* Code that inserts preferences in the System Preferences pane
* Applications that require a G4 or G5 processor
* Applications that depend on one or more kernel extensions
* Kernel extensions
* Bundled Java applications or Java applications with JNI libraries that can’t be translated

The porting guide makes an interesting read.

If this fails they always have iPod. Oh wait, didn't Bill Gates predict the demise of the iPod?

Apple's adoption of Intel doesn't change a thing vis-a-vis building Mac clones. You can't build a Mac clone, PowerPC or Intel, without the Mac ROM, which the OS checks for, and which is copyrighted.

PowerPC chips are readily available for purchase, but you can't use them to make a clone without the ROM.

People do make clones today for their personal use, but they need to get a surplus ROM. There is a market for these from old Macs.

There was a company that was making clones using used ROMs at one point ... I can' remember is they just went belly up, or if Apple found some way to put them out of business.

Mac ROMs don't exist in the new models, or even in any Mac sold in the past three or four years. They were phased out back in the days of the OS 9 to OS X transition. That's why new Macs can't run OS 9 natively (OS 9 runs in an emulation layer that includes a software ROM).

Macs use Open Firmware for booting. No ROMs, no BIOS, just Open Firmware.

That alone makes your standard PC unable to run OS X86. They just don't boot in the same way. A hack could probably be made, but that's another subject.

So what if OS X86 is made to run on white box PCs? No large shop is going to convert everyone to an unsupported solution that may break at the next Software Update.

Hobbyist types poking at their home machine do not make or break Apple in terms of hardware sales.

Open firmware is history on Intel Macs, BTW.

Gary, Apple's Intel-architecture machines do not presently have Open Firmware in them. The news coming out of Apple today — which might change in a month — is that the shipping machines will not use Open Firmware.

I don't really see what that has to do with anything, to tell you the truth. All computers have a three-stage boot process: PROM, loader, kernel. There's a PROM which handles power-on self-tests and initialization of the first CPU and its caches. That loads a very small program, usually into s-cache on CPU 0. That loader program in turn boots the OS kernel. On a Mac, the PROM is the BootROM (which includes Open Firmware) and the loader is /System/Library/CoreServices/BootX. Open Firmware hands off to BootX which launches mach_kernel. And mach_kernel is just a program.

So the firmware probably won't have any affect on anything. BootX will have been rewritten or replaced, because it depends on the Open Firmware client interface, but bootstrap loaders are unbelievably simple things compared to whole operating systems, so that's no big deal. Point is, whatever Apple uses to replace Open Firmware and BootX should have no impact on the kernel whatsoever.

(BootX, incidentally, is the program that's responsible for drawing the grey Apple-logo screen that appears when you first turn on your Mac. When the "Starting Mac OS X" window appears, that's the kernel.)

I gotta agree with Jeff on this one. Nobody seriously believes that Intel Macs are going to use stock chips, do they? Just because that's been the platform for skunkworks and developer pre-release does not mean that the production version of Intel OS X will run on it. I'm betting that Intel will deliver a custom chip that is just different enough that OS X will be able to tell whether it's running on Mac hardware.

That's all Apple needs to preserve their closed platform experience. Repeat after me:

1. Apple does not want to be Dell
2. Apple does not want to be Microsoft
3. Apple wants to be Apple - but with faster processors for their Powerbooks.

Having just gotten around to watching the keynote video — hell of a day — I noticed something conspicuous. Throughout the presentation, Steve was very careful to talk about how Apple will ship Macs with Intel processors in them. Not "Pentium" processors, not "x86" processors, not even "IA-32" processors. He said or referred to the name "Pentium" exactly twice that I counted, once when he pulled down the "About this Mac" screen on his demo computer and once when he described the transition kit for developers.

My bet? Intel is going to roll out a new line of microprocessors that are instruction-set compatible with the Pentium 4 with MMX/SSE/SSE3/whatever they're calling it this week, but that that line will not be available for sale on the open market and that it will not carry the Pentium brand name.

After all, if Apple shipped Intel CPUs in all their Macs today, they'd be something like Intel's third largest customer after (if I remember right) Dell and HP. So the idea that Intel will ship a special OEM bundle just for Apple is hardly far-fetched.

Of course, this morning at five minutes to air time I would have put money on the proposition that Apple had contracted with Intel to build PowerPC microprocessors for them. So what the hell do I know.

jeff, after reading the blogosphere blue today on this topic along with the mac-geek sites, your post has to be the most clear i've read.

the equation apple+intel=windows/mac emulation/mac clones/demise of apple is pretty much absurdity.

i could care less if the mac os is run by feeding cheeto-eating rabbits. i just want it to run, to be put together intuitively, to grow, and packaged in the same sleek boxes that have kicked the sh*t out of any pc for a decade.

and yes, it's worth the extra money. people forget that people do pay more for good design: it's not always about the dollar.

the proof is in three letters: b. m. w.

mm.

I've been discussing this with some people, and it might be related to the Xbox2/PS3 using PowerPC chips. It might have come down to who Apple wanted to behind in line for processors- Dell and HP, or Sony and Microsoft.

As far as this is concerned:

I wouldn't be surpised if some enterprising soul is able to get some subset of OS X running on a white-box Intel system.

It's called Darwin.

Dave says the nicest things. Though I think we have a difference of opinion on our animal metaphors. You picked cheeto-eating rabbits; in my post on the subject, I speculated about a hamster with an abacus. Truly we must settle this matter like men. Shall it be flintlocks at dawn, then?

I'm at WWDC as we speak, and I was in the Mac OS X State of the Union directly following Steve's Keynote address. Apple did something very smart there: They took the whole time, usually devoted to them saying "Make great apps! Here's a bunch of apps that we love, make more like them!" to saying "Here's *how* you make more great apps. Here's how you can easily, or with more effort, recompile your apps to work with the Brave New World we sprang on you last hour." They walked people through the interface rules Apple wants you to follow, and showed with examples how a number of existing apps can be recompiled to work with the Intel processors with zero changes to their code. Admittedly, these were hand-picked examples of Cocoa apps, but I didn't think they'd have any examples that would recompile to run on both PowerPC and Intel with zero changes to the actual code.

Apple also showed a matrix of the four categories of Mac OS X applications, and projected how much work it would take to fix each of them:

Scripts, widgets, Java apps - Nothing needed to fix them, they just run. These all rely on the OS for their functionality, so if the OS is Universal (Apple's new label for something that will run on both PowerPC and Intel), the script/widget/Java will be able to run same as always.

Unix applications - need to be recompiled using XCode 2.01 or higher.

Cocoa apps - May need some tweaks, then recompiled using XCode 2.01 or higher. Projected time for average Cocoa application - 2 days.

Carbon applications - More work to tweak than a Cocoa app, then recompiled using XCode 2.01 or higher. Projected time for average Carbon application - 2 weeks.

Applications written in Metrowerks - Need to be migrated over to XCode 2.01 or higher, then tweaked and recompiled. No projected time because this is an unknown depending on migrating the code to XCode 2.01 or higher.

They also frankly addressed some of the issues they'd run into with their own work, documented that in their development documentation and stressed that developers should think Universal going forward and that Apple would be working with them over the next couple of years to make this transition as seamless as possible.

TO: Will Collier
RE: Nothing Endures Like Change

This IS going to be 'interesting'. And, as with all things, there are advantages and disadvantages to every position one can take.

I think one of the primary objectives is to go after Microsoft's OS market.

We'll see....

Regards,

Chuck(le)

Having worked at the former Motorola Semiconductor Sector long enough to know they're all a bunch of idiots, all I can say is HA-HA-HA-HA-HA-HA-HA!!!

Do PC hobbyists actually build their own motherboards?

This is a fascinating move by Apple, though I don't think any of us can truly predict how it will shake out. Apple is a very unpredictable company and their true motive (i.e. what market they are tackling) will never be revealed unless they somehow are successsful at attacking said market.

Those who think Mac is gonna pull a Sega and end up a software shop are missing something that has happend in the PC industry the last couple of years. While Dell is the big boy in terms of the mass market, several boutique computer outfits have taken off (Alienware being the biggest, and Voodoo being the most "buzz" inducing). As have some hardware manufactueres making case/motherboard combos aimed at the LAN party and Home Theater crowds. Mac could do fine in this arena. I don't know the numbers but I'd bet good money that Alienware's user base is in the ballpark of Apples. People keep saying Apple doesn't want to be Dell. No doubt, but I think they could certainly be another Alienware, though geared more towards the feng shui crowd.

At my gaming weblog we've been discussing this, since my partner in crime there is a reformed Mac head. He notes that the heart of the Mac User experience is the OS, but the style is pretty important to. I see no reason why Mac can't still sell stylish computers with their OS at the heart.

As Jeff has noted, they'll likely be using a customized IA-32 architecture, so maybe a Mac will still qualify as "not a PC." Personally, I think that would be rather dumb, from a cost standpoint. If they are still customizing so much of the computer in the Mac/Intel configuration, then they'll continue pricing themselves out of the market.

The mac blog at Wired.com notes that this may be about DRM, which Intel is building into their motherboards. This alone may be enough to prevent mass hackery and keep the Mac OS paired to a particular hardware config.

One last thing regarding this mention of IA-32. If that's the case, then this move is world class stupid. Let me see if I understand this correctly. Apple spent most of the last couple of years stressing that they were the first 64-bit desktop (when in fact AMD probably beat them by a hair), and just as the x86 architecture is moving to 64-bit, Apple is going to buy into a customized 32-bit Intel processor? Does that make any since? I don't think so, and while I love to rail on Steve Jobs, I don't think even he is that stupid, particularly when Apple loves to bill itself as a cutting edge tech company. I'm not a hardware architect or anything so maybe I'm missing something, but Windows boxes are going to see a jump in performance with the move to 64-bit. I have hard time believing that Apple is going backwards, regardless of how much raw horsepower they pick up.

It WILL be interesting and fun for tech geeks to watch.

TO: All
RE: One Immediate Impact of This Announcement....

...I am NOT buying anymore major hardware NOR applications until the dust settles, probably around September 2006.

I HAD been considering purchase of VectorWorks w/Architect, LandMark, Spotlight, Mechanical and RenderWorks modules [$2000+]. But that's being put on hold as of THIS MOMENT. I was also considering updating my Mac OS Server software to Tiger [~$1000]. That's on hold as of THIS MOMENT as well.

Hope Apple and it's dedicated third-party software developers are prepared for this sort of cash flow problem.

Regards,

Chuck(le)
P.S. I think Motorola must have done something to really torque Apple off.

P.P.S. If the Intel chip boxes become overshadowed by a better chip, Apple could always move their OS to another chip; whatever may be better at the time. It's not like they've staked their life on Motorola or Intell. They're flexible. Which means we need to be flexible too.

I can't see Apple running on customized Intel chips. To undergo this effort and not be able to take advantage of the cutting edge Intel products the instant they come out makes no sense to me whatsoever. They'd also miss out on the economy of scale discounts.

But then again, what do I know?

Hey, Russ? Keep something in mind. Nobody but AMD gives a shit about 64-bit computing.

(Now that I've said something outrageous, I will back it up with fact and opinion, thereby distinguishing me from your average reporter.)

There is one and only one difference between 64-bit computing and 32-bit computing: the amount of memory that can be addressed by a single task at one time. On a 32-bit computer, a single task can address two gigabytes of memory. On a 64-bit computer, that limit is erased; a 64-bit task can address thousands of times more memory than any single computer, even a massive, multimillion-dollar supercomputer, can hold today.

But is this a good thing or a bad thing? If you need to address more than two gigabytes of memory, of course, 64-bit computing is a very good thing indeed. But who needs more than 2 GB of RAM? For the most part, only scientific and technical users need that much RAM. If you want to run BLAST or MSC.NASTRAN, more than 2 GB of RAM is basically a necessity. If you run something like Oracle, having more than 2 GB of RAM can be really helpful; more RAM means more caching, which means faster database responses. But to the average desktop or laptop user, 64-bit computing offers no benefits whatsoever. In fact, it brings liabilities.

I'm not going to write 5,000 words about this; instead, I'll try to keep it short. When you do 64-bit computing, you use memory addresses that are twice as large as the ones you use when you do 32-bit computing. As a result, you can only store half as many of those addresses in the various caches in the computer. Similarly, you can only move half as many of them through the computer's address bus in a given amount of time. When you do 64-bit computing, you're basically cutting your computer's bus and caches in half.

Net result: The exact same program compiled for 32-bit computing and 64-bit computing will, on the same hardware, run faster in 32-bit mode than in 64-bit mode. You can test this yourself today, on your G5. I first learned of the phenomenon back in the MIPS R10000 days in the mid-90s. God, that was ten years ago now. Seems like yesterday.

Now, that said, as I mentioned some people need 64-bit computing, and are happy to live with the trade-off that it brings. For these people, I'm sure Apple will have a 64-bit alternative at all times, whether PowerPC- or Intel-based. Right now, that's the G5-based Xserve and Power Mac. (The iMac doesn't really count, because you can't squeeze more than 2 GB of RAM into it anyway, so the 64-bit capabilities of the G5 bring no advantages to that computer at all.) Jobs was very specific in today's announcement to say that we'd start seeing systems in the summer of 2006, then see higher-end systems by the summer of 2007. Of course he was referring to Intel's delivery of an IA-based microprocessor that's fully ISA-compatible with IA-32 but that has 64-bit addressing. I think the Pentium 4 Prescott qualifies, but I'm not certain about that at all. The details of exactly what Intel is going to sell Apple, of course, are presumably only known to Intel and Apple right now.

So the short answer is no, Apple is not taking a step backwards from the G5 with 64-bit capabilities to an IA-32-based system that lacks 64-bit capabilities, because the first Macs to get IA-32 chips will be ones that can't take more than 2 GB of RAM anyway.

(This comment was basically copied-and-pasted from a comment I made on my own blog post on this subject. I'm lazy.)

Sorry, Jeff, but I may have to have to delete that post. References to MSC.NASTRAN cause me to vomit uncontrollably (little incident my senior year, you understand).

Don't let it happen again.

Sorry, Will. Didn't mean to be so insensitive. For future reference, how do you react to the name "ABAQUS?"

(Best to duck just in case.)

Thanks for all the info, Jeff. I hadn't thought of it, but it sure makes sense that the CPUs won't be vanilla Pentiums.

Something else to put into the mix: as Jeff suggests, faster isn't always better. Speed and memory addressing for their own sakes don't add much value to a typical user. The real problem with the G5 is that it is basically a supercomputer processor, designed to run fast in multi-processor installations where cooling or power are not issues. The deal went sour because IBM couldn't make a laptop version of it, just like you can't make a minivan version of a Corvette.

A solid 32 bit 4GHz processor that will fit in a laptop and run for 6 hours on a battery charge - without toasting your family jewels - is the holy grail of portable computing at the moment. And since laptops now outsell desktops, this market is way too large to ignore.

So worrying about power is wasting attention on the wrong thing. Apple is going for the real prize. Too bad they didn't aim for it a few years ago.

But in a sense they did, DC; didn't Jobs say they'd had OSX running for years on Intels? Now they've got Rosetta working and the iPod cash to sit on and carry them through the migration period...hopefully.

Good point by Mr. Bingley. I've also wondered if the G5 wasn't just a strategic pause while they got ready for the shift announced yesterday. It makes sense from that point of view. They likely weren't ready to leave the PowerPC platform in 2002 but they are now.

It's too bad in some ways, but the IBM vision of the PowerPC future never really matched any PC vendor's. I thought that would change, but no mobile G5 after all this time says it all.

Yeah, and this will hopefully allow them to get the graphic and memory bus systems up to snuff...but I admit I don't even care that much about that, as all of my gaming is done on consoles these days.

Sigh. As a VMS user, this just has me worried. I'm staring down the barrel of an Itanium transition in a world that is consolidating around x86.

The whole point of the Alphacide was to transition to the next industry-standard architecture, which was going to be Itanium.

Now, not only is Itanium a niche player, but everything else that isn't x86 is disappearing.

I'm beginning to suspect that Alpha may outlast Itanium, after all.

"A strategic pause." Great turn of phrase, DC. I'm totally stealing that.

I'm no microprocessor expert, but I really think the PowerPC architecture has a lot of potential. It's just that it's really expensive to make new microprocessors. We're hitting a point of diminishing returns with microchips; every unit of value we get out (performance per unit time, performance per unit power consumed, units of power consumed per unit time, whatever) requires a bigger investment of cash up front.

A company like Intel, that makes sixty-four skrillion microchips a year, can afford to invest time and money into producing what DC called the holy grail: A low-power, high-performance microprocessor. Low power consumption in a laptop means longer battery life. Low power consumption on the desktop means you can cram more of them into the same space. Imagine a "Power Mac G6" circa January 2008 with eight Intel-made microprocessors each running at 4.6 GHz and dissipating nine watts each.

IBM built stealth bombers: Spectacular devices that did amazing things, that cost a fortune, and that required you to build a special hangar to enclose them. Apple needs stealth bombers, but they also need B-52s, F/A-18s and infantrymen with M-4s. Apple needed more options, and IBM didn't have the resources to provide those options. Bless 'em.

TO: All
RE: One Immediate Impact of This Announcement....

...I am NOT buying anymore major hardware NOR applications until the dust settles, probably around September 2006.

Dunno about that... My Pismo is getting a bit long in the tooth, and I'm going to be keeping a close eye on what kind of deals Apple is offering for discontinued G4 powerbooks. And I might pick up a Mini to play with (eventually, Apple will do something that precludes installing the latest rev of OSX on my 7500....).

I was also considering updating my Mac OS Server software to Tiger [~$1000]. That's on hold as of THIS MOMENT as well.

I've played with it. Unless it does something you need, I suggest skipping it and waiting for Leopard.

P.S. I think Motorola must have done something to really torque Apple off.

Damn right they did.

Remember the G4/500Mhz debacle? D'ya think it's a coincidence that the G5 is an IBM design?

Now IBM has burned Steve (3 Ghz in 12 months!), and it's Intel's turn. If they can't deliver on their promises, there's still AMD.

i'm kinda surprised that Jobs didn't go with AMD to begin with. you'd think that Jobs, who styles himself as a free-thinking mavrick, whould have gone with AMD since they have the precieved reputation as the Intel alternative.

and i'll admit it right now: i love AMD. i've been using variations of the Athlon for years now and will probably use Athlon 64 4000 in my new game rig.

if and/or when Apple finally releases just an OS that will run on an intel/amd box, i will be all over it like a rottweiler on a beef pot roast.

Steve doesn't care nearly as much about the image of Apple's suppliers as their ability to deliver the goods.

Oh, and you'll be a long time waiting for that version of OS X that runs on a 'white box' PC. Apple's hardware sales is what pays for the operating system R&D. No hardware sales, no operating system R&D.

Am I the only one that thinks that PC users have been staring at mac hardware for years now, wishing that one of the clone makers would make mac copycat hardware that was even half as good as what apple was supplying? I don't think that the interesting problem is in running OS X on a beige box intel PC. I think the interesting thing will be to make windows run on Mactel hardware. Suddenly, all those PC geeks that were giving their money Dell, Voodoo, and the others will be considering the Mactel hardware at or near the top of their list. As pointed out in the original post, apple really like their hardware business due to its high margins, and even their PPC hardware with its much smaller production volumes, is price competitive on a feature for feature basis with pc clones, so I don't really see a problem (other than getting windows to run on it). Sure, Dell offers MORE price points than apple, but where they make similar hardware, they also have similar pricing, at least within a margin that is tolerable for the kind of consumers that cares about the style and architecture of their computer hardware.

Half of apple's customers are repeat customers, so there is no threat that those folks will be migrating away from OS X just because they can use windows. Meanwhile, some of those wintel folks who are considering purchasing mac hardware are surely going to leave a dual boot setup, since it will come with os x, regardless. OS X is such a superior desktop computing environment (IMNSHO), that I can't help but think that mactel hardware which is capable of running windows is likely to generate switchers TO the mac operating system, since they will now have much easier access to apple software. Surely there will be a pearpc/vmware/transitive style virtual environment for running os x at near native speeds from right within windows (and vice versa). Regardless, how many gamers have you met who will say things like "I'd love to run a mac, but I can't play games..." A dual-boot mactel computer solves that problem. Just boot into windows for gaming, os x for everything else. Hell, we did the same thing for years with MS-DOS mode in win95 and 98.

Basically, what I am getting at is that this decision maybe takes hardware consideration out of software platform choice for many users (many of the deep pocketed ones), and given the high regard with which apple hardware is generally held among the geekier and trendier set, won't that probably be an upside for apple's hardware business in the long run?

I'm an idiot when it comes to anything beyond actually using my mac. But although I know that I'm considered a heretic for mentioning it (believe me, my more fundamentalist mac friends have told me so), I think Apple has to create a design that allows PC software to run. As mentioned before, the biggest complaint I hear from users about Mac is the dearth of software choices.

And as much as purists scoff at this, it is the fundamental reason that Mac remains at 3% (or whatever the current market share is). I've stuck with Mac for almost 20 years now and I wish they would finally start taking my humble but brilliant advice.

BB, we can stop spewing nonsense about market share now. That metric is fundamentally specious. When you cite installed-base numbers, like your "3%" figure, you're comparing Apple to all other computers everywhere. That's not a useful comparison.

There are two metrics that are useful. One is total installed base. That's in the neighborhood of 40 million, according to IDC. IDC estimates that, of all the Macs that have ever been sold ever, 40 million of them are in use today. Something like 90% of those run Mac OS X; 65% run either 10.3 or 10.4. That's a huge market for software vendors. The Mac has never suffered for lack of software, and it's not going to start any time soon.

Second is share of sales. Apple is the third-largest personal computer maker in the world, behind Dell and HP. And it's a close race for second place. If Apple's CPU sales rate of growth for the second quarter of this year were to be equal to its rate of growth for Q1 — 40%, year over year — Apple would challenge HP for the #2 spot in the world.

When you say, "Oh, only 3% of computers are Macs," you mislead by omission. Only 9% of computers are Dells, and Dell is the #1 vendor on the planet. So total-market-share numbers are both meaningless and deceptive.

Apple is a much bigger, much healthier company than many people seem to realize.

"And then it's all over for Apple as a hardware vendor."

Hacking OSX onto other hardware will be done by a small but technologically sophisticated group of people. Because these are people who wouldn't buy Mac hardware or software anyway - it's not a lost sale for Apple when small groups of hackers do this. Overall it will be such a small and insignificant number of people that it will not affect Mac sales.

"Hacking" Mac OS X on to other hardware won't happen at all. Google "LaGrande."

It does become a problem when someone bundles up open-source hackery into a product, as has happened with WINE. "Buy a Mac OS X retail box and our $50 package and run it on your PC!" DRM will be the only way Apple will be able to prevent it.

I'm frankly more worried about being able to run Windows programs on the Mac at near-native speed. Either WINE or Microsoft's virtualization tech (acquired from Connectix; they had Virtual PC for Windows) could be used for this and Microsoft has so much to win by letting Windows apps run on Mac OS X, I'd be shocked if they didn't revive the product. Adobe, Microsoft, and other big software shops would love to be able to ship a single version of their apps, and if Windows apps can run reasonably on the Mac, I predict we'll see it by the time the first Intel-based Mac ships. Then with everyone using Windows versions of their major apps anyway, it becomes easier to think about switching to Windows to save a few bucks at the next hardware upgrade cycle, because frankly, Windows apps aren't that bad. I can think of a number of ways to handle the UI differences between Win and Mac so that Windows apps would look decent in Aqua.

Now this would also make it easier for Windows users to switch to Mac (they could run all their big apps verbatim), and this may be Steve's plan, but it's risky. Apple is, sadly, no longer the UI leader it once was. I use both Mac OS X and Windows XP on a daily basis and while I bought a G5 because I love it, I have to admit that Windows XP just is about as usable and reliable on quality hardware. It's not as pretty, but it can be made so; Microsoft just needs to hire some competent graphic designers. The only solid thing that the Mac currently has going is the virus/spyware thing, and I don't see that as a sustainable competitive advantage ("moat"), since Microsoft has plenty of money to throw at the problem and has recently been glove-slapped by none other than Intel's CEO.

I sure would like to see Apple pull it off, and I personally will probably buy a 2007 Macintel model, but going head-to-head with Microsoft in this way is dangerous. So far they've won every time and it would be dumb to bet against them.

I'm wondering if the custom ia32 variant being discussed is based on the PIII core like the pentium-m chips. Not only would this be a relatively low power design it would also have a superior IPC and lower memory bandwidth requirements, probably equivalent to current Athlon models.

It's true that the average desktop user does not see any direct benefit from upping the word size from 32 to 64 bits the X86-64 is a completely new architecture with a completely overhauled architecture with double the number of registers.