OSX on my PC?

June 6, 2005 12:07 PM

Everyone's seen the news that Apple's moving to Intel processors, right? Does this mean that I will finally be able to install OS X on my PC?
posted by redteam to computers & internet (21 comments total)

Probably not. Even if OS X ends up running on a strict x86 architecture, instead of Itanium or something, it will very, very likely use Openfirmware rather than PC Bios at the least. And there will probably be a few other proprietary elements.

What is more likely is that it will be very easy to run Windows without any performance hit on your Intel-based Mac. Again, assuming processor architecture compatibility.
posted by weston at 12:19 PM on June 6, 2005

It don't see why it would as Apple used to run IBM processors and that didn't have any impact on using OSX on an IBM computer. I may really be showing ignorance here, but the hardward shouldn't have any impact on the software that's run on it, no? Apple will still be running Apple machine language on the processors, not PC machine language. Or do I have no idea what I'm talking about?
posted by spicynuts at 12:23 PM on June 6, 2005

a few other proprietary elements

Um, replace proprietary with "different." Open Firmware is not in fact proprietary.
posted by weston at 12:24 PM on June 6, 2005

Probably not. We're not even sure that shipping Apple/Intel machines will be running current P4 (etc.) processors. The OS X/Intel 10.4.1 preview package that is available starting today runs on a P4, but who knows what the shipping specs will be.

The main thing is the BIOS, the thing you see on the PC when the computer boots up and you see lots of unattractive text (really, really sorry if you're, like, a PC expert...heh). Apple will probably use a different one than common PCs use. There will be a bunch of little tweaks like that that will (probably) lock OS X off the generic computers.

On the other hand, this could be the next logical phase for Apple: getting out of being primarily a computer company, and focusing on their consumer electronics, home entertainment, and similar lines, which are giving them enough cash to survive the inevitable hit in hardware sales that today's announcement will cause.

What I want to know is what this means for the whole 64-bit move... That's why I'm wondering what processor OS X/Intel will ultimately run on.

On preview: what weston said more concisely.
posted by socratic at 12:26 PM on June 6, 2005

weston: AFAIK, OS X will be ported to emt64 (aka amd64). Even then, that's just the processor -- peer around at the rest of the parts inside your PC, and remember that Apple would have to support everything in there (and yes, especially the BIOS) for OS X to run on a generic amd64 box. Apple has a strong marketing and financial interest in preventing OS X from running on other manufacturers' hardware and in preventing Windows from running on their hardware, so expect them to enforce same.

spicynut: Machine language is, well, the language of the machine, in this case the processor. Windows on emt64 and OS X on emt64 will both use the instructions emt64 processors use (but see above). The PowerPC platform in the Mac and in IBM's Unix workstations had a great deal in common, enough that Linux and NetBSD and friends could share a lot between the two ports.
posted by mendel at 12:30 PM on June 6, 2005

Unfortunately a processor is just part of the equation. The entire motherboard architecture plays a huge role, as in how devices are managed by and communicate with the process, as mentioned before, the bios, etc. It would probably not be *impossible* to get OSX to run on something other than the current Mac architecture -- it's very common outside the Windows world to have operating systems that run on multiple architectures. This is generally accomplished by writing the operating system in something like C and deliberately adding as little platform dependant code as possible, and relegating all of the arch dependant stuff to the periphery of the OS (device drivers and the like). Apple does not have to do this, they can effectively make their OS run only on the architectures they want it to. And they probably will.
posted by RustyBrooks at 12:31 PM on June 6, 2005

Well, there is a Darwin build that can run on x86. I imagine that if Apple doesn't make an official build to run on generic (read: not Apple-branded) x86 hardware, somebody out there will make one.

On a related note: With regard to the running Windows on an Intel-based Mac thing, weston - will this finally end the whole "which OS is faster" debate? The current processing speed comparisons always are criticized for comparing (PowerPC) Apples to (x86) oranges.

On preview, spicynuts: From what I understand the issue here is primarily the way that the OS communicates with the hardware. PowerPC processors have more open sockets, so communicating with the processor is easier - more channels to send info through. Drivers will have to be recompiled to address the lack of sockets in an x86 chip. The lack of x86-native drivers is why PearPC for example runs so slowly - it has to emulate the Darwin core environment, and translate calls to hardware from Apple mode to Intel mode on the fly. Right now there are no Apple drivers for x86 chips. If Apple makes some available, there might not be any reason why installing OSX on your Dell wouldn't work. You used to be able to home-build your own Mac. It would be nice if Apple brought this back in the move to Intel, even if officially unsupported. A big part of my reluctance to move to a Mac is that I don't like giving up my control over the hardware. I like to build my own PC, which you can't currently do with a Mac. Building your own means you aren't locked into the original company when it comes time to replace defective or outdated hardware. Building your own Mac would also make it easier for the average DIY PC enthusiast to give the OS a shot - lower barrier to entry if the hardware is cheaper and more readily available.

RustyBrooks - also one of the big reasons why Windows is more unstable. MS had to build it to run off any old pile of parts, while Apple has total control over what platform configurations are available. Much easier to do real-world testing for stability when the hardware you test on is the exact same hardware your customers will be using.
posted by caution live frogs at 12:33 PM on June 6, 2005

Here's some paranoid speculation for you. You know all those rumors about Intel activating DRM in their new processors? The main purpose of that DRM could be to restrict execution of OS X to those processors that contain Apple-specific secret keys.
posted by Galvatron at 12:37 PM on June 6, 2005

Weston pretty much said it all... One of the Apple VPs said afterwards that they will not allow OS X to run on non-Apple hardware. He also said that people will likely run windows on Apple machines but Apple wouldn't support it.

So, save your pennies for a Pentium Apple. It looks like x86 and not Itanium will be used. I guess I'll be saving up for that Pentium M PowerBook to come in a year or two... never thought I'd be saying that.
posted by gazole at 12:38 PM on June 6, 2005

There won't be an official, supported retail version of OS X for PCs as a result of this announcement, no. (Hacks are a whole 'nother issue.)
posted by mcwetboy at 12:41 PM on June 6, 2005

here is my 2¢....
Apple continues to make and sell hardware bacause they have a ton o' cash tied up in those little things called The Apple Stores. and that said hardware will rely on not open firmware but actual ROM chips (remember those?) so no OS X on the $500 Dell machine.

what do you think?
posted by ShawnString at 1:19 PM on June 6, 2005

I think Galvatron hit the nail on the head. If Apple really wanted to improve power consumption/performance by switching processor architectures they would have gone with AMD.

IBM is perfectly capable of producing a high performance low power chip. Why they are not doing so is beyond me, except that IBM is now supplying a lot of chips to Microsoft...
posted by wezelboy at 3:28 PM on June 6, 2005

IBM is capable of doing it, but they might not have a business reason to do so. IBM doesn't sell G5 based machines, desktop, notebook or other form factor. The G5's major purchaser is Apple so essentially they're in a position where Apple would need to pay most of the R&D costs. IBM is more interested in making high performance IBM Power series processors, some of which eventually trickles down and helps out the G series.

With Apple going to Intel along with most of the rest of the world they get the performance increases for free since Intel is driven to maintain Moore's law.

AMD is the wrong move. In the long term Intel will crush them. Apple has already been burned twice by not switching to Intel.
posted by substrate at 4:23 PM on June 6, 2005

I prognosticate: There will be no OSX intended for use on commodity PCs. However, it will not take much effort to hack it to do so for certain hardware configurations. Apple has been creating and testing x86 builds for a while, so obviously it runs (however well or badly) on some subset of commodity hardware as of this moment.

It is patently obvious that the target instruction set is amd64 (a.k.a. x86-64). Itanium is less than worthless for real world processing loads, and x86-32 would, in more ways than one, represent a step backwards for Apple.
posted by majick at 4:55 PM on June 6, 2005

fwiw, I heard on Tech TV today that next-gen Apple OS's will be able to be installed on generic x86 pc's.
posted by ttrendel at 7:11 PM on June 6, 2005

majick, you're talking out of your ass about Itanium's worth. It's a very respectable processor for scientific computing. It's the wrong processor for a desktop or even a workstation though.
posted by substrate at 7:25 PM on June 6, 2005

It will be possible to run OSX on a PC, but it will not be easy, it will not be legal, and it will not be seamless. You'll no-doubt get a small sub-culture of hackers who do it for the sake of doing it, but it won't be something that the average person can do. My 2c :)
posted by -harlequin- at 7:27 PM on June 6, 2005

Apple is going with Intel over IBM because IBM has been unable and/or unwilling to deliver improvements on a schedule that keeps pace with the larger PC industry. The fact that they will be delivering faster chips for game consoles is irrelevant, except to underscore their lack of commitment to PowerPCs suitable for Mac desktops and notebooks.

Apple is going with Intel over AMD for two likely reasons:
1) power consumption to performance ratios of Intel's Centrino mobile processors is better than that of AMDs
2) Intel offers just about everything, processors, chipsets, BIOS, compilers, video, network chips, etc and no doubt have the resources to make the hardware transition as easy as possible for Apple. They also tend to be ahead of AMD at introducing instruction-set enhancments for multimedia data types

That said, it should be relatively easy, from a technical standpoint, for Apple to move to AMD in the future.

The reason Apple isn't going to let you run it on any old PC hardware, is first and foremost, that Apple is a hardware company, not because of its investment in Apple Stores. If MacOS ran on third party hardware, like Dell, Apple would see their hardware margins evaporate and its unlikely that they would be able to grow their operating system business fast enough to compensate for the lost profits from hardware. (Which isn't to say that it won't happen, someday)

OS X on generic Intel would also dilute the Apple reputation for design and ease of use that comes from controlling the software and the hardware.

As for this:

PowerPC processors have more open sockets, so communicating with the processor is easier - more channels to send info through. Drivers will have to be recompiled to address the lack of sockets in an x86 chip.

Perhaps "more open sockets"="more general purpose registers," which is true, and explains some of the theoretical performance advantages the PowerPC has over intel, expecially historically. The fact that it doesn't seem to matter much illustrates the huge economic advantage Intel has by having a larger market share in a much larger hardware market.

The availability of general purpose registers doesn't have a whole lot to do with the ability to support a broad range of PC hardware. Modern computer operating systems, and even device drivers, are written in C or C++ and don't contain much assembly code and compilers generally handle register assignments.

It will be interesting to see what people come up with for running OS X on non-apple intel hardware. Darwin would seem to address many device driver issues, though probably not with the 3D accelerated video which has found its way into the basic UI.
posted by Good Brain at 7:30 PM on June 6, 2005

fwiw, I heard on Tech TV today that next-gen Apple OS's will be able to be installed on generic x86 pc's.

And this is why you shouldn't pay attention to anything on Tech TV. It's total nonsense.
posted by xil at 10:50 PM on June 6, 2005

I've seen it mentioned several times that Apple will need to support the multitude of PC motherboards, BIOSes, etc.
This is, IMO, nonsense - Apple could (and should) come up with a "minimum required specification" MacOS x86 motherboard on which the OS is guaranteed to run, and for which all of the drivers are provided. Anything else would naturally need to be caveat emptor.

This was, in effect, the idea behind the CHRP ("Common Hardware Reference Platform") that was driven by Apple, IBM, and Motorola - albeit from a PowerPC point of view. It's also a similar way that the new AmigaOS has developed - it's PPC, but it is only guaranteed to run on specific hardware.
posted by Chunder at 5:44 AM on June 7, 2005

"majick, you're talking out of your ass about Itanium's worth."

Let me repeat my statement, this time with emphasis: "Itanium is less than worthless for real world processing loads..."

Yes, it's a pretty decent architecture for straight-line integer loads and floating point performance is not at all shabby, however a vast majority of extant code in the world is not straight-line integer. It appears to be widely accepted that the state of the VLIW compiler art is currently insufficient to compensate for that.

So, um, no, I'm not talking out of my ass. I just happen to know what non-niche, general-purpose processing loads of compiler-generated code actually look like, and have seen firsthand how astoundingly miserable the Itanium is at running them. Scientific computing is a small fraction of total CPU cycles expended in the world; it does not and should not drive designs for general purpose computing.

There is zero chance that an Itanium -- or any mostly-pure VLIW procesor -- will appear as the primary execution unit in a future Macintosh.

Assuming this move is successful, expect Apple to stick with the amd64 architecture for years to come.
posted by majick at 1:48 PM on June 8, 2005