Slashdot Mirror
OS X on x86?
FusionJunky asks: "There has been some talk surrounding Apple's OS X and its potential to be released for the x86 platform. Sites like OSXonIntel.com have been trying to get the message to Apple that we feel the consumers are ready to see OS X on x86 boxes. I'm wondering what the Slashdot community thinks this would do to Apple, would it adversely affect their hardware sales? Could Apple move away from selling G4s from Motorola and start producing Intel Macs. Do you think Apple should release an x86 version of their next gen OS?" We asked earlier whether you felt if Linux would be threatened by OS X, with the possibility of OS X working on x86 machines, has your answer changed?
Let's think about this.
If they're running on commodity x86 hardware then hardware sales would go through the floor. Who would buy the expensive (but good) Apple hardware when there was cheaper hardware which did the same job? Doesn't matter if the Apple stuff is better, the man in PC World isn't going to know that so he buys the cheapest. Apple lose out bigtime.
If they're running on their own dedicated x86 hardware (remember here, there's a difference between having an x86 CPU and being IBM PC compatible) then they have to develop that, so they've got a new one-off cost to develop this, along with the same costs to maintain it as they currently have with PPC - except they don't have experience with it so they'd be running slower for a little while. Oh, and they'd lose the current advantage of not having fans, as the x86 CPUs run a lot warmer as a whole than PPC. I can't see they'd go to all this trouble so they could use laptop or Crusoe processors. They'd also then have to move their software over to the new CPU for the second time in recent memory, write all the converters and so on. So they wouldn't have a speed advantage for a couple of years (think upgrade cycles) as it'd be emulated. Not pretty.
All this so they can use different CPUs which have a current speed advantage (which the change would knock out for a while). Could someone possibly explain why this could be a good idea?
Greg
(Inside a nuclear plant)
Aaaarrrggh! Run! The canary has mutated!
Whilst I would openly welcome OSX on my desktop, running on my x86 hardware with all the funky hardware I've bought I see this as the one major problem for Apple. If OSX were to be released for the x86 market the number of configurations and hardware devices the OS would have to be able to support would grow to the sort of levels which would be problematic. At the moment they only have to support a few graphics chips out of the box - how many would they have to support if they were attempting to sell OSX to the masses. What about soundcard, modems, network cards and so on. I just can't see it happening without a lot of support from manufacturers who are unwilling to provide driver support on the existing mac platform.
If OS X for x86 was available, I'd be willing to bet that a large proportion of the first two groups would immediately jump ship over to cheaper hardware. Mac OS hardware is nice (The G4 is a great chip) but being realistic, price is king.
The question, therefore, is whether the increased revenue in OS sales would compensate for the losses in hardware revenues. Personally, I'm not sure.
When it comes down to it, though, Apple won't listen to what we say. Pretty much every major decision will depend on what Steve thinks, and I don't think Steve likes the idea of open hardware - historically, he's shown himself to be very sensitive to physical appearance (the original mac classic, the iMac, the cube) and I suspect that he just doesn't want his lovely OS X running on ugly grey boxes. :)
(Cue announcement from Apple tomorrow about OS X for x86, just to prove me wrong...)
For those of you who insist on purporting that Apple makes all of it's money on hardware, I suggest you take a different look at it. My question to you is, how much money would Apple actually make selling hardware, if that hardware didn't come with the Mac OS attached to it? Would you buy an overpriced piece of hardware just because it looked cool, but had to run windows on it?
If your answer is no (as I suspect it is) then the reason you buy Apples is primarily because of the software and OS, which makes Apple primarily a software vendor who uses their software technology to sell hardware. Without the Mac OS, the Mac is just an overpriced PC encased in dayglo plastic. Sure some of the hardware touches are nice, but none of that really matters without the Mac OS.
Apple is primarily a software vendor people. Don't forget that. Just because their accounting puts most of the revenue in hardware sales doesn't mean that hardware is truly where their competitive advantage is.
I've heard this tripe for so long, it makes me ill....
"No, no, no. Don't tug on that. You never know what it might be attached to."
In my humble opinion, Apple would release OSX on x86 hardware on only one condition. That it was *Apple* x86 hardware and not everything under the sun that is currently available for an x86 based platform.
... Cocoa? ... will be the easiest to port. As I understand it Cocoa is a subsystem API that talks to the Mach microkernel, not to the hardware. So Apple would port the Cocoa API to x86 with the rest of the OS and the apps that use it would port easily with a few changes, just like OpenGL games do now. The "hardware" apps would obviously be the hardest, and would cause the most harm to Apple's partner alliance. Take Adobe for instance. They have worked very hard to get their Photoshop for the Mac to take direct advantage of the AltaVec Velocity Engine instruction core in the new G4 chips. There will be no Velocity in an AMD chip, so that entire section of their code would have to be thrown out, or rewritten. Being that they already have a Winport and a RISC Macport, doing a specific x86 Macport might be just enough to cause those companies using Velocity specific code to jump ship entirely.
Meaning that the issues that people have mentioned about Legacy problems with the x86 hardware would not exist. The situation would go something like this. Not that I ever see this happening, but it is the only way I can think of to make it work.
Steve would decide that Apple needs a good, powerful sub $500 system to compete with the likes of Dell, Compaq, Gateway and eMachines. The Apple Design Department would start working on an Apple standard platform mainboard (meaning that this would not be an ATX/AT box, it would be a G4/iMac box) that uses only *one* of the current x86 processor technologies. They would choose between either Intel or AMD and stay there.
Let's assume the choice would be AMD, because of cost per processor and the fact that in my scenerio they want to beat out the $499 eMachine AMD based box. The mainboard would be designed to be PCI *only* with USB as the same keyboard / mouse input we see on current PowerMacs. (Don't worry about issues with ISA slots, Apple would not write drivers for anything in them anyway.) This means the mainboard would be similar to the "legacy-free" Compaq system, the iPaq. No serial, no parallel, no PS/2 ports. Just USB and some PCI slots. Oh, and don't forget the single AGP port for graphics.
This indicates, that even though in theory Apple is porting OSX to x86, they are really porting OSX to AMD. Though Intel is also an x86 chip, they do have different command sets than the AMD chips have and vice versa. OSX (in my scenerio) is optimized for the AMD processor, so most likely a good number of the options of the OS would not work, or simply break entirely, if run on an Intel processor.
Also, there would not be drivers for everything under the sun, including the breadboard PCI card your Grandmother built to keep her recipe system organized. Apple would choose a small amount of cards to build drivers for itself. (The cards that ship with the system) Then it would be fully up to the hardware manufacturers themselves to write drivers for their hardware. If it isn't "Apple Certified" you can bet that Apple won't even give you a deaf ear to voice complaints upon.
Also, software that runs on OSX will need to be rewritten for most apps. Apps that are specifically coded in
Basically it looks like Apple would be shooting themselves in the foot to port to x86 at this stage in the game. The only reason to do it would be to have a sub $500 price point system. They would have to retool entire manufacture specs on a board specific to x86. They would have to port the OS, and all of their specific drivers to a platform that would *still* only work in the Apple way. (They aren't going to give me technical support on OSX if it is installed on a Gateway PC.) Then they would have to train technical support to field questions on the new x86 based box.
Kind of like SGI and their current market position. I never thought I would see the day when I could purchase an x86 box running NT Workstation / Linux with an SGI logo on the front of it. But I can, and I still have to ask myself "Why?" As far as I know, SGI still has some MIPS based hardware running IRIX for sale. Yet they also have these x86 boxes that they have to develop, troubleshoot, support and market. These SGI boxes are no more or less special than an IBM or a Dell system with the same specs. Apple is not about to let *anyone* say that about them.
No, I don't think OSX will have be fully available on an x86 box until long after Apple is dead and gone as a company and someone has bought all of the intellectual rights just for the purpose of porting OSX to x86 hardware. It isn't that it is impossible. The core of the system is already running on x86 (Darwin) and could probably even get as crazy as NetBSD or whatever and be running on Dreamcast, and Amiga, and VAX, and Commodore, and my toaster oven. The kernel is portable, it's just not going to be ported. It just doesn't make a good business sense looking at the current Apple business model.
"Genius may shine aloof and alone, like a star, but goodness is social, and it takes two men and God to make a Brother."
While Linux and the various flavours of BSD all currently offer solid environments for x86 hardware, the desktop facilities offered by all of them are still poor in comparison to most commercial opertaing systems.
Despite the hype surrounding some of them, I've not found any Unix desktop environments particularly compelling so far... There's also plenty of hype surrounding GNOME and KDE, but I don't see any killer applications for them. People don't need buzzwords, they need a powerful desktop environment running on Unix, and this is what Apple will be providing.
Linux and the BSD derivatives are excellent server environments, as is Darwin, but MacOS X is also getting Maya and the Adobe applications on a viable desktop.
Personally, I've yet to find a desktop environment that offers me everything that I want, BeOS came close, but Apple are in a position to offer everything in one package.
Apple are also in an excellent position to pick up the gauntlet dropped by SGI, right at the top end of the content creation market. 3D, publishing, design and video editing are about to get a whole lot better on the Mac.
The most likely reason Apple will want to continue development of Darwin for x86 is that it needs a backup plan, exit strategy if things go wrong.
If Motorola and the G4 architecture don't scale fast enough, if Apple can't buy the hardware in the volumes they need, if the price of PPC processors is too high, then Apple would still be able to put x86 processors into it's systems with only minor re-engineering OS X.
Custom designed Apple x86 hardware is therefore a possibility, and an exciting one, but I doubt Apple see any commercial interest in supporting generic PCs and legacy hardware.
Also remember that OS X has the useful ability to support application "bundles" with binaries for multiple platforms in a single executable. This would make it very easy for vendors to ship applications in a form that would run on both PPC and x86 transparently to the user.
-- aardvark doesn't constitute a good password.
Now even if Apple decided to make x86 boxes of their own, and make them right, i.e. with all the little features and the good stability and support we've come to expect, Apple would still be in deep trouble. They would release a version of Mac OS X that only runs on Apple's x86 hardware, of course. But the core, Darwin, is open-sourced, so it would be a matter of days (or maybe weeks) until someone comes up with drivers to let it run on other hardware. We've seen how much trouble Apple had with clone makers, now think about what would happen if Apple moved to x86...
Anyway, Mac OS X would have a hard time on Intel. If you're used to Linux, compiling and running Unix software on Mac OS X is (excuse the wording) a pain in the ass. I know what I talk about, because I ported GNOME to Mac OS X (see http://fink.sourceforge.net/). Also, there is no Classic environment on Intel. This is a big drawback, because not all applications will be ported to Mac OS X right away, and you won't have the choice of running the classic Mac OS applications of x86 hardware.
Personally, I'll just keep running Mac OS X on my PowerBook G3 and be happy...
-chrisp
"If that makes any sense to you, you have a big problem."