Apple to Switch to Intel CPUs?

[ewmayer]

Originally Posted by rogue

Personally, I don't see this increasing market share. Who says that Apple wouldn't ask a premium for a Intel-based Mac? Most of their sales come from hardware, not software.

Well, it's hard to separate HW/SW with Apple - IMO they've used distinctive branding roughly equally on the HW side (PPC/AltiVec) as the SW/OS side (Mac OS X). But I don't see how they *can* expect to charge a hefty premium for an Intel-CPU-based solution - they won't well be able to argue that their CPUs are technologically superior to Intel's anymore, will they? (Unless they've got some 64-bit PPC/Itanium-style trick up their sleeve - more on that below). So their only remaining distinction would seem to be on the software/OS side, and there a sellable premium would appear to be limited to the $100-200-per-CPU range, basically what one expects to pay for an OS/SW bundle.

I wonder if they use the P4 as-is or if they will use a special version of the P4. I also wonder about 64-bit support. Some of us (granted a distinct minority that I am in) use the 64-bit instruction set on the G5 writing assembler. Will I have to learn x86-asm? AFAIK, Itanium has been not been successful. Will Apple use it or does Intel have something else up their sleeve?

Well, if you think about it, leaving aside the CISC-vs-RISC issue (which is of course sweeping an elephant under the rug, but let's assume we have a big rug for the moment), the hardware capabilities of the G5 and the Itanium are really quite similar in many key ways. Owing to the relative failure of the Itanium I could see Intel desiring a G5-like architecture as a low-price solution for the Itanium's target market ... which makes it all the more bizarre that Apple would be abandoning the PPC. So unless Apple/Intel do have a joint PPC/Itanium-style solution planned for the high-end market, I really question the wisdom of this strategy - as I see it, Apple has been successful in roughly equal measure to the quality of their product as to the cachet of being unique and different. And let's face it, now that Microsoft has got their act together to the extent that most of the hatred for Windows is political/dogmatic rather than based on objective quality-of-product standards, I'm guessing the latter (i.e. "I'm unique and special, a thinker-outside-the-box, a non-runner-with-the-herd," etc.) is a bigger source of revenue for Apple than the former. In that light, this seems like a really questionable business decision. But maybe for the typical Mac aficionado, love truly is blind...

[cheesehead]

Perhaps moo has found a clue on Tom's Hardware Guide as to why Apple is switching?

http://www.mersenneforum.org/showthread.php?t=4167

[rogue]

The main problem in porting from x86 to PPC (besides assembler routines) is not the language, but the APIs. I don't see any real advantage in porting Windows software to Mac since the APIs are so different. IF (and that's a might big IF) Apple allowed OS X to run on x86 boxes built by Dell, HP, etc., then existing companies could use existing hardware to run OS X without having to buy new hardware. This might make development on OS X more favorable.

Another issue I've seen with Tiger is that the Cocoa framework does not support the 64-bitness of the G5. I wonder if Apple will not do that work now that they have laid the path for Intel.

What is also interesting is that Tiger was just released a few weeks ago and it does support 64-bit addressing and 64-bit ints (in console apps). The timing of this announcement implies that they intend to support Intel's 64-bit CPU out of the box.

What is also interesting is that it appears that they will compete directly with Longhorn in some markets.

BTW, I just found this interesting link:

http://www.wired.com/news/mac/0,2125...w=wn_tophead_2

[ewmayer]

Originally Posted by rogue

BTW, I just found this interesting link:

http://www.wired.com/news/mac/0,2125...w=wn_tophead_2

So if this is to be believed (and hey, it's certainly more plausible from a bottom-line perspective than anything else I've seen), Apple has gone 100% Hollywood. No problem with that, just as long as they drop any pretense of being different from any other money-rules-you-tools corporation. (Somehow I don't think they will...)

Sing along with me now: "We're only in it for the money..."

Another Zappa lyric also comes to mind:

"Your mind is totally controlled,
You have been stuffed into my mold,
And you you will do as you are told,
Until the rights to you are sold..."

[dsouza123]

Does this mean there will eventually be a OS X version of mprime ?
OS X is based on BSD and there is a FreeBSD version of mprime.

Issues :
32 or 64 bit executable,
ability to use object file format used by both Prime95 and mprime 32-bit,
ability to use object file format used by Prime95 64-bit.

[mephisto]

Originally Posted by ewmayer

I don't see how they *can* expect to charge a hefty premium for an Intel-CPU-based solution - they won't well be able to argue that their CPUs are technologically superior to Intel's anymore, will they?

Nope. But how many people in the past 10 years have actually bought Apple hardware because of 'technical superiority'? For market share, look for supports what I need to do (which for Apple is graphic design, video and music production, and the mandatory support for Word and Excel), good word on the street (approximately that people you know use it for the same thing you are going to use it for, and are happy with it) and coolness.

I see no problem in Apple separating SW and HW, selling the SW as an alternative to Windows and the HW on the design. Apple knows design in a way that PC makers have never managed, even 20+ years into the PC standard.
Think from a consumer perspective: There are all these boxes at approximately similar prices and approximately similar capabilities, you might as well pick the best-looking one of the lot.

I have no idea of what Apple is actually thinking or doing, but there are several possible scenarios that make perfectly sense.

Originally Posted by ewmayer

So their only remaining distinction would seem to be on the software/OS side, and there a sellable premium would appear to be limited to the $100-200-per-CPU range, basically what one expects to pay for an OS/SW bundle.

On the other hand, by separating the OS and HW they might increase their OS sales tenfold.
I've liked the OS X operating since it came. Still, I never got around to buying an Apple, just because the investment is so large (HW + SW), and because other factors (work) requires that I have a capable computer that can run Windows. Of course I could buy two computers, but the 'somewhat cooler' advantage of OS X never outweighted the extra cost and bother of having two separate systems. But if OS X came for X86, I would be first (well, at least probably among the first million or so) in line to buy it. I know plenty of people that run Linux and Windows on the same box, I expect there are quite a few that might want OS X and Windows as well.

There are problems with such a move, of course. The largest is probably that basically all existing software for Mac has to be rewritten. Not a small task, to be sure :)

Still, I like the move: Finally, a GUH (Grand Unified Hardware) has emerged for consumer-level computers. About time, I think. Alternative architectures will find their niches elsewhere, in PDAs, gaming consoles, graphic cards and whatnot, so the competitive aspect is still there - too many false moves by Intel, and the Cell CPU might eventually take over.

[rogue]

Originally Posted by mephisto

I've liked the OS X operating since it came. Still, I never got around to buying an Apple, just because the investment is so large (HW + SW), and because other factors (work) requires that I have a capable computer that can run Windows. Of course I could buy two computers, but the 'somewhat cooler' advantage of OS X never outweighted the extra cost and bother of having two separate systems. But if OS X came for X86, I would be first (well, at least probably among the first million or so) in line to buy it. I know plenty of people that run Linux and Windows on the same box, I expect there are quite a few that might want OS X and Windows as well.

There are problems with such a move, of course. The largest is probably that basically all existing software for Mac has to be rewritten. Not a small task, to be sure :)

Mac OS X will only run on Apple hardware. They have already stated as much. Will it be possible for someone to hack it to make it run on non-Apple hardware? I don't know, but I don't expect it would be easy.

Regarding the rewrite of software, your are wrong. The only software that has to be rewritten is software that has assembler routines and software that uses AltiVec. That is a small percentage of all software available for the Mac. The API will be the same. Apple already has a version of Xcode that can compile fat binaries, i.e. binaries that can run on both PPC and x86. They also have Rosetta which will run PPC binaries from within OS X on x86.

[ndpowell]

Originally Posted by xilman

Note that Intel !== x86 ...

Paul

That could also be Intel := x86, or Intel == x86, or just Intel = x86. Just wanted to make sure that the others weren't left out

[Dresdenboy]

Originally Posted by ndpowell

That could also be Intel := x86, or Intel == x86, or just Intel = x86. Just wanted to make sure that the others weren't left out

Didn't he mean "Intel != x86"? How about "Intel == (IA32 || AMD64 || IA64 || ARM || i960)". These should be the most important ones

[Peter Nelson]

Originally Posted by Dresdenboy

Didn't he mean "Intel != x86"? How about "Intel == (IA32 || AMD64 || IA64 || ARM || i960)". These should be the most important ones

I think Intel would disapprove that you incorrectly include AMD64. They would instead say you meant to say EM64T (which is almost, but not exactly the same as AMD's architecture). ROFL.

With the amount of plugging of dualcore and EM64T on desktop from Intel, these are strongly likely to be features of a cpu used by Apple.

One reason is look at the expected release dates for Apple's Intel machines, and compare with Intel processor roadmap. I think a kindof dualcore Pentium M is due out shortly before and could therefore get used by Apple. I think we ought to assume that EM64T will be the supported instruction set, and in 64 bit mode.