Mac OS X on pentium and amd

[jermsmingy]

before you flame me consider this. I don't think that apple will allow clones again, but since from what I hear mac os x is either mostly ported or can easily be ported to wintel. What if apple did that but the only way that you could run os x on a pentuim or amd is if you bought the computer from them. so you could buy a quicksilver with a g4, pentium 4 or and amd. Now, granted they could only ship the the ported version of mac os x with these computers to guarentee that people would not install it on a wintel box, but I think apple could do it. Then no one could say that apple is slower or behind, because people could use the processor of their choice. Also, this wouldn't hurt apple's hardware sales.

Just wondering what you guys thought.

[ftaok]

Let me start by saying that I don't think this would happen.

But here are some points that I think would be interesting to discuss.

1. Would OS X on a 2 Ghz P4 be any faster than on a Dual 1 Ghz G4? If not, then switching to Intel/AMD would not buy anything of substance. I read something in the "mainstream" media about the PPC platform being a lot faster than Intel/AMD for Linux, even taking into account the clock speed differences.

2. If Apple does go this route, I expect that the motherboards would be designed by Apple. That way, there wouldn't be any clones. If Apple were to go with a standard Intel motherboard with a Mac ROM installed, how easy would it be for some techie to "clone" a Mac ROM?

3. How happy would developers of Mac software be to learn that they would have to port their Mac software over to the Intel platform while still keeping the PPC version? If Apple were to switch to Intel, they would have to drop PPC. No "straddling" on this one.

[idkew]

as interesting as this could be, it will NEVER fly.

first, it would pose too many problems trying to get all software to run correctly on two VERY different archetectures.

apple would obviously have to double its hardware development, 2 processors, 2 motherboards......

the gap is not wide enough to warrant such an investment in "new" technology (to apple). if you know enough about the ppc vs pentium, the pantium is rarely faster mhz to mhz and most of the time it is doing the same process a ppc does in one clock cycle in 2 clock cycles. that is the reason for the astronomical clock speed. the pentium 3 is much much better in this case.

the only reason i see for a switch is to get moto and ibm off their asses and get some germanium (sp?) and soi going real fast. or apple could just buy the semicunductor division from moto.

[strider42]

Originally posted by idkew
or apple could just buy the semicunductor division from moto.

Don't people realize that apple could never afford the semi conductor division of moto, nor would moto ever want to sell it. apple MIGHT buy out moto's interest in the aim alliance, as they have a right to do (or something like, I'm not totally clear on the details), but the semi conductor division would probably be worth billions.

Other than that, I agree: running OS X on intel would probably not serve much of a purpose, and would only complicate things.

[blakespot]

Originally posted by idkew
as interesting as this could be, it will NEVER fly.

first, it would pose too many problems trying to get all software to run correctly on two VERY different archetectures.

apple would obviously have to double its hardware development, 2 processors, 2 motherboards......

the gap is not wide enough to warrant such an investment in "new" technology (to apple). if you know enough about the ppc vs pentium, the pantium is rarely faster mhz to mhz and most of the time it is doing the same process a ppc does in one clock cycle in 2 clock cycles. that is the reason for the astronomical clock speed. the pentium 3 is much much better in this case.

the only reason i see for a switch is to get moto and ibm off their asses and get some germanium (sp?) and soi going real fast. or apple could just buy the semicunductor division from moto.
I think it's only if Motorola dropped the ball (well, farther than they have) on the PowerPC that Apple would go x86, but a proprietary system architecture (you'd need Apple hardware). But as for the scenario of supporting multiple platforms with one OS---driver issues would be the biggest hurdle. Let's remember that OS X started its life as NeXTSTEP on the 68030/040-based NeXT machines. It was then ported to x86 and eventually became ported as OpenStep to Sun's as well as HP machines. The "fat" binaries that floated around contained code enough to run on Intel, Motorola, Sun, and HP machines. There were differences in stability though. HP's implementation was not nearly as stable a the original Motorola platform, which is the most stable, but not as fast as some of the Intel hardware, etc.

It can be done--but it's a headache. But remember---OS X did not, by any means, start its life on PowerPC.


blakespot

[AlphaTech]

I think that it would take way too long to port OS X to the wintel platforms. Consider how many different pieces of hardware are out there for the peeecee. You have several brands of video cards as well as sound and network cards. That alone could delay it for months. Either that, or Apple would have to say that it would only run on certain configurations. With specific motherboards as well as other cards (for sound, ethernet and such) creating the drivers could be a task that Apple just doesn't want to tackle at present. If m$ ever does go the way of the dodo bird (as we all are praying for), then Apple would need to address all the peecee people that get left in the dark (OS wise).

[firewire2001]

"I think that it would take way too long to port OS X to the wintel platforms"

Well think of it this way... if you look at any other release of UNIX or LINUX, they have moderate compatiblity, but almost always full compatibility with video cards, sound cards, and keyboards and mouses - the essentials. These drivers are all open source, and wouldnt be hard at all for apple to port... for instance, apple still uses many of the same printing packages/drivers that mainstream UNIX uses today...

[alex_ant]

I think if Apple were to port to Intel (which they would not do), they should have the decency to send the following letter to all current registered Mac owners:


Dear PowerPC Macintosh owners,

**** you.

Sincerely,
(hand-written signature)
Steve Jobs
President and C.E.O.,
Apple Computer

[AlphaTech]

I still think that the chances of Apple porting OS X to the wintel platforms has about as much chance of a blizzard happening in Miami, FL. Yes, it could happen, but not in our lifetime, or this millenium. :D

Mac's rule, peecee's just drool

[pishnaris]

The NeXT OS, which was by far the best OS ever developed, and is the parent of our beloved OSX, was ported to Intel processors...it never made it. NeXT was just plain unbeatable, hardware and software-wise. Unfortunately, it was WAY TOO expensive, and the 68040 at 33mhz was as far as it went.

But what a machine it was in 1990!

[Catfish_Man]

Originally posted by strider42


Don't people realize that apple could never afford the semi conductor division of moto, nor would moto ever want to sell it. apple MIGHT buy out moto's interest in the aim alliance, as they have a right to do (or something like, I'm not totally clear on the details), but the semi conductor division would probably be worth billions.

Other than that, I agree: running OS X on intel would probably not serve much of a purpose, and would only complicate things.

'Fraid you're wrong about Apple not being able to buy Moto's semiconductor division. It's in their contract, they can buy it for (if I remember correctly) 500 million. 500 mil is a big chunk of cash, but Apple could theoretically afford it.

[rEd Eye]

Imagine being able to choose the hardware you wanted at an affordable(key word "competitive") price,whilst using a superior os.
I don't think that there is a chance in hell that Apple would give anyone this chance,because no one would bother with Apple$ hardware anymore!
My guess is that soon enough M$ or whomever will copycat Apples idea of revamping unix into a prefab consumert os,that's just as good,if not better than os X.Then the ***** will hit the fan,because Apple won't be better anymore!
Although I am a mac user,I am obviously not a fanatic,as you can probably tell.I find the Apple os more functional for my purposes,therfore I have no choice but to buy their hardware.Given half a chance,I would happily build myself a tailored to my needs hardware setup,and slam os 9,and os X into it,leaving the Apple prefab kit to those whom are looking for a second BMW to go picnicing on Sundays with.

(Anyone ever noticed how os X strangely resembles Win XP?:D)

[b8rtm8nn]

I wouldn't be surprised if OSX Server would be ported to a different platform in a couple of years, but only the server. That would mean a venture of an IBM or HP providing high end server hardware and drivers that are labeled Apple, and it would probaby work is OSX Server keeps maturing at the rate it is. ever happen for the desktop/laptops though.

[ftaok]

Originally posted by Catfish_Man


'Fraid you're wrong about Apple not being able to buy Moto's semiconductor division. It's in their contract, they can buy it for (if I remember correctly) 500 million. 500 mil is a big chunk of cash, but Apple could theoretically afford it. Catfish,

I'm afraid that you're confusing Motorola's semiconductor division with their PPC interests. MOT's semiconductor division is huge and while Apple has $4 bln, they're not prepared to spend it all for a division that, while has high revenues, has low profits (if any).

The $500 mln that you're talking about sounds like MOT's PPC interest only (although I've never heard of it --- doesn't mean it's not true). $500 mln wouldn't be enough to buy the entire semiconductor division.

But I guess it's just semantics.

[jbouklas]

If you remember the NeXT years, Steve Jobs released NeXT for the NeXT boxes, Intel, and SPARC. He built the binaries of the OS in such a way that a program written for the Intel platform would work on any of the others. This is what OS X needs to be a giant. If we had OS X on Intel, Mac, and Sparc, with cross-platform binaries, then one piece of software would work on any hardware platform. This would be tremendously useful to Apple, as they could have iMac workstations, a couple of Transmeta-based servers, and whatever communicating with Intel boxes all over the world. It would be the perfect corporate tool. But, M$ should be long dead first.

-Jim

[pishnaris]

My recall could easily be faulty, but I don't believe a NeXTStep app, written for the 68040 would run on an Intel or Sparc machine...the app still had to be compiled for the other platform, although that wasn't a terribly difficult thing to do.

[teabgs]

Originally posted by rEd Eye


(Anyone ever noticed how os X strangely resembles Win XP?:D)

I hate to break it to you man, but OSX came out BEFORE XP.

Win XP looks strangely like OSX....just like every other version of windows looks similar to mac OS.

[jermsmingy]

my whole point was what if apple but a pentium or amd in their hardware.
PowerMac pentium IV. I would buy the g4 still but I think it would reach a different part of the market. My only question is, I am not a developer, but you all keep saying that if it ran on a different chip, then developers would have to maintain different versions of the software. Why is that. I can understand if their program accesses hardware, but os x doesn't let any program do that. It accesses the hardware and then lets the programs use it. Someone please explain.

[sparkleytone]

close this thread, im tired of this talk.

[jbouklas]

This is a magazine clipping from the good old days. Outlines what I've been saying:NeXT OS (http://next.z80.org/img/next_nextstep/page5.jpg)

[ftaok]

Originally posted by jermsmingy
my whole point was what if apple but a pentium or amd in their hardware.
PowerMac pentium IV. I would buy the g4 still but I think it would reach a different part of the market. My only question is, I am not a developer, but you all keep saying that if it ran on a different chip, then developers would have to maintain different versions of the software. Why is that. I can understand if their program accesses hardware, but os x doesn't let any program do that. It accesses the hardware and then lets the programs use it. Someone please explain. I don't think that we missed the point at all. While I believe that it's possible that Apple could get OS X to work on an Intel/Apple hybrid machine, I don't think that said machine would be any faster than what's currently provided with the Dual G4. I'm not sure about this, just speculation based on what was said about the PPC being a better architecture for Linux.

And for Apple to switch to Intel, I think that they would have to drop PPC altogether as the different processor lines would cause confusion in MacLand. I can see it now, harried salespeople trying to explain why a 1 Ghz G4 Power Mac costs more than a 1.7 Ghz P4 Power Mac. "But these go to 11."

[PCUser]

A program is written in a high-level lanugage (like C, C++, etc) and is compiled (exculding, of course, things like Java, but it can't make a very good stand-alone Word processor) to Assembly.

Assembly is a processor-line-specific language. The PPC processor line has it's version of the Assembly language, and the x86 has it's own. They are incompatible.

Once the code is turned into Aseembly, it still needs to be able to call system libraries that are dynamic (ie, they aren't part of the compiled code, such as system calls to make windows and buttons, etc), so they are linked to those libraries. Once everything is linked together by a program called a linker, it is handed to an assembler program that turns the Assembly code into machine language.

Machine language is processor-line-specific, as well. Apps assembled for PPC will not work on 68k or x86. Apps written for x86 will not work on 68k or PPC.

So, it's not to do with talking to the hardware calls. That isn't the issue. The issue is that source code turns into processor specific Assembly (Motorola has different ideas about what should go into Assembly then what Intel and AMD think), is linked to system libraries that are compiled as well, and then turned into machine code.


Yes, you can emulate the other chip architecture (ala VirtualPC, etc), but it will never be as fast as the actual chip itself is at comparable speeds.

Also, the system libraries must be generated for x86 if you want MacOS X apps to work on x86. Right now, those libraries are written and compiled with PPC in mind. Quartz, Cocoa, Classic, etc are written and compiled for the PPC architecture. Darwin, the core of MacOS X, works on x86. But no GUI-level libraries do.

[jbouklas]

How did they do it with NeXT Step? It was advertised that one application could run on several hardware platforms. How did they do it?

-Jim

[Taft]

Think of this...

Apple is currently competing with Dell. One of Apple's best weapons in this competition is its ability to differentiate in its OS and software. Now lets port all of that software to Intel. In my opinion that takes away a great weapon.

Now its possible that Apple moves away from directly competing with hardware vendors. But as of right now, they make most of their money from hardware. Taking away the ability to sell that hardware would kill them.

Matthew

[PCUser]

Well, I don't know how NeXT did it, but the only thing I could think of would be to make it so the system libraries handled every call...

Such as basic calls to add, divide, subtract, access memory, etc. The whole works. You'd have to recreate the entire Assembly language in the system libraries (no small feat).

You'd also have to write your own compiler, linker, assembler so that it compiles your code to call the system libraries for everything.

That would be a lot of time to do. And, I imagine, it would a slower then if you wrote it the standard way. It'd basically add another layer to everything. Right now, when I write a simple C++ app and compile it, it talks directly to the chip to make calls to add two numbers... if I was to recreate a system-independent system library to do the same thing, my app would get compiled to call the system library to talk to the chip to add two numbers.

[strider42]

Originally posted by Catfish_Man


'Fraid you're wrong about Apple not being able to buy Moto's semiconductor division. It's in their contract, they can buy it for (if I remember correctly) 500 million. 500 mil is a big chunk of cash, but Apple could theoretically afford it.

Sorry, you are not correct. apple has the right to buy out moto's interest in the powerPC (or it might just be int he aim alliance), NOT the semiconductor division. the fab plants alone would cost more than 500 million. 500 million is not that big a chunk of change when you are talking abuot somethign as vast as moto's semiconductor division. Its not possible. apple could never buy it, and have no right to do so. Whats more, what would they want it for. the semi conductor division does a lot more than just apple chipsets, meaning apple would get saddled with a whole bunch of stuff it doesn't want thats not even profitable.

[swahilibill]

I believe that MS does need OS X, hell even Mac OS in general, To make some real software, but that is like trying to put a V8 in a 87' Oldsmobile, it just does not look or work right.

[alex_ant]

Originally posted by PCUser
Well, I don't know how NeXT did it, but the only thing I could think of would be to make it so the system libraries handled every call...

Such as basic calls to add, divide, subtract, access memory, etc. The whole works. You'd have to recreate the entire Assembly language in the system libraries (no small feat).

You'd also have to write your own compiler, linker, assembler so that it compiles your code to call the system libraries for everything.

That would be a lot of time to do. And, I imagine, it would a slower then if you wrote it the standard way. It'd basically add another layer to everything. Right now, when I write a simple C++ app and compile it, it talks directly to the chip to make calls to add two numbers... if I was to recreate a system-independent system library to do the same thing, my app would get compiled to call the system library to talk to the chip to add two numbers.
I must say, when it comes to using one's own hind quarters as an information resource, you are more skilled than most. :)

Fat binaries work because they are nothing more complex than several different binaries joined together into one executable. There's no translating going on - each section of the binary executes native code respective to the processor on which it is being run. Any fat binary will run on any system for which a part of it has been compiled, and of course the binary will use up at least twice as much disk space as a thin binary depending on how many targets it has been compiled for. Fat binaries are no great technological feat.

As for porting to Intel: I cannot imagine any way Apple could do this successfully. Even assuming their machines remained proprietary, with a boot ROM for example, I wonder how long it would take for some bored 16 year-old kid from Hong Kong to work around that and get OS X running on all Intel machines, legal or not. Probably about three weeks.

Why would Apple WANT to take the huge step of moving to Intel, anyway? Sure 1GHz < 2.4GHz. But Apple is doing quite well and making money. Where is their incentive to switch?

I can think of one company that parallels Apple in a way: SGI. In 1997, they were doing quite well, but the MIPS R10000 was starting to fall behind Intel in terms of megahertz. Nevermind that the R10K was superior to the Pentium II in virtually every way; it was at 225MHz and Intel was at 300MHz, and that hurt SGI's sales. So SGI decided to spin off MIPS, embrace Linux, embrace Intel, and port everything on over. They helped port Linux to Itanium, and were in the process of porting IRIX as well. In the meantime, they sold proprietary x86 boxes running Linux and NT. SGI's x86 boxes completely flopped. They still sell Itanium servers running Linux, but those aren't selling either. In short, when they bowed to pressure and abandoned the great IRIX/MIPS platform, they shot themselves in the heart. SGI is now a mere shadow of its former self; it has re-embraced IRIX/MIPS, but even they know that it's too late. They would likely no longer be in business today if not for their classified sales to the U.S. Government.

Everybody who wants their next Power *MACINTOSH* to include a 2.4GHz Pentium 4 with a 15,000 RPM CPU fan sucking up god knows how many watts of power and re-establishing a tired and miserable legacy architecture that should have been dead ten if not fifteen years ago, raise your hand.

Everybody who wants to essentially kick EVERY SINGLE current Mac owner in the nuts by declaring the PowerPC a "legacy architecture" as Old World Macs are now, and by abandoning approximately 90% of the current Mac-using population (because face it - this is what would have to happen if OS X were to be ported to Intel), raise your hand.

Everybody who wants to alienate current major players on the Mac scene such as Adobe, Macromedia, Microsoft, etc., who make the software that makes the Mac a viable platform, by forcing them to again update the software they've just spent so much effort getting to run on OS X, raise your hand.

Everybody who wants to see all the hard work and investment Apple has put into the PowerPC architecture over the past decade thrown by the wayside in favor of a temporary and short-sighted solution to a problem that frankly is not so enormous anyway, raise your hand.

In short, everybody who wants to see the Mac platform die a slow and agonizing death, raise your hand.

Please - can the Intel talk already.

Alex

[alex_ant]

Originally posted by rEd Eye
My guess is that soon enough M$ or whomever will copycat Apples idea of revamping unix into a prefab consumert os,that's just as good,if not better than os X.Then the ***** will hit the fan,because Apple won't be better anymore!
Well, you could say Be already did this, or was at least close, although the only thing the "*****" hit was the face of Jean-Louis Gassee and all BeOS users.

Alex

[PCUser]

I don't think Microsoft would ever switch to a UNIX-clone based OS... simply because that would mean they would have to admit that Window's kernel wasn't the best crap on earth. They wouldn't even have to say it, just switching to a UNIX-clone kernel would admit it.


Alex_ant, thanks for setting the record straight about NeXT. You learn something new every day!

:) I did not intend for anyone to think I knew what I was talking about, that's why I prefaced it with a "I think" disclaimer. I was guessing, not stating as fact. I don't think I deserved that. :(

I was thinking that NeXT used a thin binary that could run on multiple platforms. If it was just a fat binary, what's so special about that? :confused: Other OS's can do that too...

[eirik]

Someone has already pointed out that Pentiums and Athlons draw a hell of a lot more power than PPCs, generating tremendous heat.

While placing a couple slots in the tops of Pentium/Athlon units so one could toast bread, muffins, or bagels might be a novel approach that would garner tremendous free publicity (mind share is proportional to market share...generally), the maintenance would be a huge inconvenience.

With my present PC's, I seldom clean the dust off my CRT's and boxes. I can't imagine ever popping the box to empty out the crumbs.

Now, to make matters worse, some brilliant marketer might convince Apple to employ a side-loading slot with convection ducting so that one could cook small pizzas. Well, I love pizza, man do I love pizza. But now we're talking about having to clean out the grease and various nasty food spatters.

No, with the increased threat of a major energy crisis due to the problems in the Mideast, we must conserve energy. To employ Pentiums/Athlons, we would be ethically obligated to go the toaster-oven route to justify the power. As Bush Sr. used to say, "it wouldn't be prudent"!

Seriously though, I do have a practical point to make that hasn't been raised in this thread, though others. Apple is clearly strengthening its position in multimedia and scientific workstations. And I suspect, later it will muscle into Internet/Intranet servers as an industrial strength, low overhead (headcount) solution. This screams for multiple processors (and vector processing).

Not only do the Intel/AMD CPU's draw a butt-load of power but they also do not satisfy all of the MERSI criteria for optimal symmetric multiprocessing. Mind you, Intel/AMD may have improved upon this but if they have I haven't heard about it yet, which wouldn't be a shocker.

With PPC's, except for the G3, being very MERSI compliant, Apple can get much better SMP performance with PPC's than with Intel/AMD CPU's, so long as the individual CPU's are not radically disparate in performance.

With 300mm wafer production propagating through the industry, the price of CPU's will go down substantially, providing greater pressure for mainstream employment of and workstations' more extensive usage (more CPU's) of SMP.

The server market is already itching for lower power CPU's for their server blades because the Intel/AMD-based hardware requires additional expensive power supplies and cooling equipment that easily rack up an additional $15,000!

Apple enjoys some positive differentiation from the rest of the market with its reputation for quieter machines because its machines draw less power and bleed less heat.

I'm not bullish on Intel/AMD right now. However, if Moto and IBM don't get their s*** together this year, I'll start growing horns.

As for the $500 MM buyout option of Motorola's PPC interests, I agree that this does NOT include the fab assets. Apple would be extremely foolish to buy those assets. Those assets are failing to keep up with Intel/AMD manufacturing advances (I am referring to mass production of high-clock CPU's, not advanced technology such as SOI, AltiVec, and copper interconnects.). Managing them would seriously distract Apple from its core competencies. Their lower margin generation would bring down Apple's overall margins. Apple's cost of capital would go up, as well as its financial beta.

The big question that I can't answer is just exactly what would such a buyout include in the way of the intellectual property that Apple relies upon? Would it include AltiVec? SOI? Copper interconnects? Silicon Germanium compositing?

If the option would get Apple the desired technologies, Apple shouldn't hesitate to execute, even if Apple continues to source Motorola in the short term. Over time, I would expect that Apple would outsource manufacturing of its in-house designs, courtesy of this option plus Apple's existing intellectual assets and human power, to AMD, IBM, Intel, and/or some Asian fabs.

Well, there I go again. I wrote another damn novel!!!

BTW, please remember to clean out your Intel/AMD based PowerMac's periodically. It might smell nice but its a fire hazzard. :D

Eirik