Yesterday IBM published a story on their Intranet promoting their Technology Group and their relationship to Apple with regard to the PowerPC 970. This article has been made available to MacRumors. The item offers some unique insight into Apple, and their relationship with IBM as well as Apple’s thoughts on the alternative, Intel.
Woot!
I think it’s interesting that Apple had considered moving to an Intel-based chip. Of course, that doesn’t necessarily mean it would have been an x86 chip since Intel could have created a whole new chip for Apple.
In the end, though, as a Mac user, I am really glad that Apple had apparently solidified a long-term relationship with IBM, who I anticipate will be a whole lot more reliable than Motorola has been. I think this Apple-IBM relationship is laying a good foundation for the long term growth of Apple.
Everything seems to be clicking just right. Damn, it’s great being a Mac user these days!
Woot!
I think it’s interesting that Apple had considered moving to an Intel-based chip. Of course, that doesn’t necessarily mean it would have been an x86 chip since Intel could have created a whole new chip for Apple.
In the end, though, as a Mac user, I am really glad that Apple had apparently solidified a long-term relationship with IBM, who I anticipate will be a whole lot more reliable than Motorola has been. I think this Apple-IBM relationship is laying a good foundation for the long term growth of Apple.
Everything seems to be clicking just right. Damn, it’s great being a Mac user these days!
I agree, moving to Intel wouldn’t necessarily mean moving to the x86, in fact, if the Itanium wasn’t so power hungry and produced so much heat, the logical step would have been to move from PowerPC to Itanium.
1) Itanium EPIC takes the ideas of RISC to the nth degree. Early compilers for RISC were shocking, however, many years later, they’re pretty good. EPIC moves more of the responsibilities to the compiler and because of this, the moving of code from RISC to EPIC shouldn’t be such a hard task and code which used many of the features on RISC should be able to be transferred with out too many issues.
2) Moving to Intel wouldn’t necessarily mean having to use the crusty x86 design for the board. Apple could easily retain the use of Open-Boot and thus simply design a chip-set and board which could accommodate the new CPU.
3) AMD would never have been a viable alternative due to the clear supply problems that AMD have. Hype a chip and only to find that in two months time they can’t keep up with the demand. They’re the x86 version of Motorola.
sure there were all the rumors, but really
You would be a fool to think Apple didn’t consider going x86, or any other cpu out there. As a company you have to weigh the options. And on so many fronts the move to something like the AMD64 solutions. But clearly as the letter states Apple felt that would be very difficult. I’m sure Marklar was part of this investigating. If it wasn’t for IBM being able to delevier a cpu to apple they wouldn’t have had much choice but to switch. Apple knew it was in it deep on the cpu front, they were very lucky to come out with something in the final hours. Apple didn’t have much time to spare or much of their base that was waiting for the G5 would have just moved on.
Of course, that doesn’t necessarily mean it would have been an x86 chip since Intel could have created a whole new chip for Apple.
Now why would they do that?
It was probably posted on IBM’s intranet for morale boosting, but serious, I find no suprise in this. Every smart company weighs all their options they’ve, and I mean all before making a decission. It’s a business fact.
“Now why would they do that?”
Why wouldn’t they? Siging an exclusive contract to provide all chips for Apple for X years might have been incentive enough. Who knows?
The only reason I’m jealous of the MAC is the efficiency of the power pc architecture. If Apple ever decided to port to Intel, it would be absolutely worthless. You are better of with the more reliable Unices (*BSD/Linux) that have been tested and trusted on the Intel architecture.
If only power pc motherboards are as freely available as x86 counterparts. *sighs*
Regards,
Mystilleef
I’d guess that the *NIX foundation of OSX would make it easy to change the chipset. How much of OSX is hand-coded assembly, and how much of it is (easy to recompile) C, C++, or underlying *NIX code? If there isn’t much processor-specific stuff going on, it shouldn’t take long. And with the foundation being *NIX I’d guess there’s not much.
Or maybe I have no idea what I am talking about.
Apple doesn’t have nearly enough market-share to warrent Intel making a chip just for them. Hell, even Motorola didn’t consider Apple big enough game to keep their G5 project going (or even keeping the G4 up-to-pace). The only reason IBM is doing the G5 for Apple is that IBM needs the chip anyway for its workstation machines. Also, AMD could easily supply Apple’s requirements. AMD has about 15-20% of the x86 market. Apple has about 2-3% of the total PC market. Supplying additional CPUs to Apple would be easily within AMD’s capability.
While I think the PPC970/G5 was a terrific choice, and I like the PPC platform, the actual processor Apple uses doesn’t matter. Here’s the thing that people miss…..
An Apple system with an Intel chip is not a PC!
Is a RS/6000 a Macintosh? How about a router that uses an embedded PPC?
Apple would do away with alot of the horrible things about the Wintel platform. It would not be able to boot DOS, or run Windows, without modifications to the way Windows works.
Still, all of this is unimportant news. It would only be news, really, if they’d been foolish enough not to consider all the options.
An Apple system with an Intel chip is not a PC!
Exactly. This is a fairly major point both the “Apple should never go intel” and “I want OSX on intel so I can run it on my home built clone” crowds don’t seem to understand.
If Apple made a machine with a P4 or other x86 CPU in it, it wouldn’t be able to run current intel OSes out of the box (although the first Linux port would probably only take a week or two).
Similarly, the intel version of OS X wouldn’t run on any hardware that wasn’t Apple’s – at least, not without some major hacking at the Darwin layer (probably take a few months for a version of Darwin to allow OS X/intel to run on any clone).
The biggest reason Apple wouldn’t have gone intel unless they absolutely had to would have been software support. Being squeezed out of the market by clone builders would barely have been an afterthought.
Apple would do away with alot of the horrible things about the Wintel platform.
Most of those “horrible things” are irrelevant to pretty much everyone except a handful of low-level developers. Apple x86 boxes would be rid of them, to be sure, but their removal would certainly offer little functional improvement to end users.
It would not be able to boot DOS, or run Windows, without modifications to the way Windows works.
It almost certainly would never be able to boot DOS, as that requires pretending the hardware is an early-80s era XT (although maybe someone could hack something together in Open Firmware to emulate it ?). Similarly for the DOS-based versions of Windows.
NT-based versions of Windows would probably only need minor modifications to the HAL to work – although I sincerely doubt Microsoft would ever bother (what would be the point).
Note that darwin comes out of the box with i386 support. Apple has a patent on multi-architecture binary , and they use it. NeXT would provide four Architecture in one binary, making /usr/bin available throught NFS.
—
http://perso.hirlimann.net/~ludo/blog/
Note that darwin comes out of the box with i386 support.
That’s a pretty generous definition of “support”. Heck, good ol’ DOS would run on more x86 hardware out of the box than Darwin does.
Apple has a patent on multi-architecture binary , and they use it. NeXT would provide four Architecture in one binary, making /usr/bin available throught NFS.
Undoubtedly, but the show stopper isn’t software support from the tools bundled with the OS, it’s from the software that all the customers want to run.
Yes, this was the obvious and reasonable thing to do, but it’s still worth pointing out that Apple’s capable of that, as most of the OS News readership (or, perhaps postership?) seems to think Apple’s run by morons and frauds
.
JT
CooCooCaChoowrote:
> AMD would never have been a viable alternative
Besides, AFAIK AMD does not build chips, just designs them and hires IBM to build them. So if IBM could custom-design a chip for Apple, why go for AMD. AMD does not seem it was an option ever.
Rumours of “Project Marklar” – being a full Mac OS X system and applications running on an Intel-compatible configuration – have been circulating for some time. A project, hidden deep within Apple’s labs, silently stalking the latest Mac OS X releases…
The suggestion that the project might be used to expand Apple’s horizons beyond the Mac, rather than in spite of it, was always there, but has elevated in stature since the release of G5 PowerMacs.
Maybe we’ll see it, someday.
Besides, AFAIK AMD does not build chips, just designs them and hires IBM to build them. So if IBM could custom-design a chip for Apple, why go for AMD. AMD does not seem it was an option ever.
AMD has its own facilities plus it outsources to a number of other foundaries. The issues of supply occured during the dot-nuts days when every man and his dog was buying a PC. Unfortunately the rapid demand for AMD chips out stripped supply. Those issues have now been resolved through the outsourcing programme plus the extra slack in the CPU production foundaries.
Also, why would IBM custom design a chip when they had a high end chip that simply needed to be scaled down for general use, both in blades, low end servers and workstations.
Rumours of “Project Marklar” – being a full Mac OS X system and applications running on an Intel-compatible configuration – have been circulating for some time. A project, hidden deep within Apple’s labs, silently stalking the latest Mac OS X releases…
The suggestion that the project might be used to expand Apple’s horizons beyond the Mac, rather than in spite of it, was always there, but has elevated in stature since the release of G5 PowerMacs.
Maybe we’ll see it, someday.
That is nothing but a load of crap. It would be like me starting a rumour that Microsoft is going to make its own UNIXWare simply based on the fact that Microsoft licensed some UNIXWare IP off SCO.
The problem is that people saw Rhapsody running on x86 then assumed that there would be MacOS X. It never happened because it would never work. x86 is far to complicated in terms of the 1000s of combinations that exist for any hardware to be supported adequately.
Microsoft is in a fortunate position where all the drivers are developed by third party vendors and are given to Microsoft to test and bundle with Windows, *NIX and any alternative operating system unfortunately has to provide all the drivers for the user because of the pig head arrogance by some hardware vendors to open up their specifications so that drivers can be written.
Dresden FAB is what then? They build opterons and athlons there (Real silicon), IBM is helping them with the next process and extra fab capacity
http://www.amd.com/us-en/Corporate/AboutAMD/0,,51_52_502_509,00.htm…
http://www-3.ibm.com/chips/news/2003/0108_amd.html
Even if the information is bogus, “Project Marklar” is a cool name. The question is whether there are enough South Park fans out there to appreciate it
Apple’s management had to consider the options available to them. That does not mean that apple was going to choose intel The paragraph in that story illustrated the situation well. The switch over would be extremely difficult for the user base and the developers.
Apple can achieve the same objectives by increasing market share, and hence volumes for the chip makers to aggressively invest in the platform. I do believe that a strategy for aggressive market share increases will show once (if) apple succeeds in diversifying its revenues base away from desktops (with servers, software, consumer electronics (ipods, work stations).
When they need a new CPU, they might have concidered the x86. They should have also looked at all the other CPUs on the market; like the MIPS, ARM, etc ,,,
The last thing Intel has said about the x86 was that the P4 is the last chip in the family. They hope to have the P4 running at about 20GHz by 2008 and discontinue it around 2010. The I2 is the future for Intel; they’ve just release the first workstation versions of the chip and by 2008, they plan to start producting PC versions to replace the P4s.
Given that Intel is killing the x86 by 2010, I don’t see why Apple would have moved from the PPC to x86. Intel’s plans for the I2 don’t happen fast enought for Apple; they couldn’t wait until 2008 for the next upgrade of the Mac family.
Most of the other CPUs on the market a designed for servers or enbedded devices. IBM anounced the development of the PPC 870 for use in it’s blades and workstations; thus, Apple did the next logical thing and used it. It’s compatable with the current 32 bit PPCs and IBM plans to upgrade it for the foreseable future.
Now, if only the x86 fans would go look at Intel’s future plans, we might get them to stop trashing alternet CPUs and start looking for a replacement. The only company currently planing to support the x86 after 2010 is AMD; and they plan on moving it to 64bits and droping a lot of backwards compatability.
Apple had early versions of MacOSX running on x86… the screenshots are still floating around the web somewhere.
> Apple would do away with alot of the horrible things about the Wintel platform.
Just commenting to say, that SGI tried this with some of their own rather proprietary versions of dual P3’s, which wasn’t hugely successful.
“I’d guess that the *NIX foundation of OSX would make it easy to change the chipset. How much of OSX is hand-coded assembly, and how much of it is (easy to recompile) C, C++, or underlying *NIX code? If there isn’t much processor-specific stuff going on, it shouldn’t take long. And with the foundation being *NIX I’d guess there’s not much. ”
According to my Cocoa Developer books not much of the kernel is written in the native assembly language. So it could be easily ported to other chipsets. I don’t know why some mac users are offended at the thought that Apple would port Mac OS X to x86. I don’t think it’s a bad idea, although you probably won’t see it sold anytime in the near future and you’d lose a lot of benifets going from PowerPC to x86.
The problem with porting Mac OS X to x86 is hardware support. I mean look at the problems Linux is having because hardware vendors won’t release their specifications so drivers can be written. Plus Apple makes more money off of hardware than anything else so they would lose a major source of income.
PowerPC processors are much more powerful than x86 processors anyway. Thats why the Navy bough a bunch of Xserves because they wanted them for the fast processing power that PowerPC processors could offer them.
Maybe Cocoa based applications are easy to port to another processor, but what about Carbon based. Carbon based programs run under OS X as well as under OS9.
A lot of software for the mac is Carbon based.
From: akumaX (IP: —.ipt.aol.com)
“PowerPC processors are much more powerful than x86 processors anyway.”
Interesting thought.
I was going to argue the point, but I see the AOL.com in the address…
Forget it…
The problem with porting Mac OS X to x86 is hardware support.
No it wouldn’t. As soon as it became known Apple was looking for hardware for their new machines, manufacturers would be falling over themselves to be the one company to provide it.
Not only that, but the only bit of hardware that would need “new” support would be the motherboard. Pretty much everything else is already the same as that used in PCs.
Apple almost certainly keeps several parallel development streams of OS X running on completely different hardware platforms like x86, just like Microsoft would with Windows and non-x86 platforms. They’d (both) be crazy not to, otherwise platform dependencies will start creeping in (which is why NT wasn’t developed on x86, but ported to it).
Hardware support is the least of the problems Apple would face moving to an x86 based platform. I doubt it would even rate much more than a footnote in a proposal document.
I mean look at the problems Linux is having because hardware vendors won’t release their specifications so drivers can be written.
The difference between a bunch of independently operating developers spread all over the world and the single legal entity of an American-based company to a hardware developer is so fundamental I’m amazed you even tried to make the comparison.
In short, Linux developers don’t get hardware specs because because a) companies are afraid of what will happen to their trade secrets if they give them to a bunch of people preaching the GPL and b) there’s little money to be made from doing so anyway. Neither of those two caveats would apply to Apple (whose only additional “x86” hardware would be a motherboard, in any event).
Plus Apple makes more money off of hardware than anything else so they would lose a major source of income.
Nothing would stop Apple keeping their same prices on proprietry Apple x86-based hardware. If anything, they’d make *more* money because they could leave their prices at the same level but the hardware would probably be cheaper for them to buy.
As someone else said, an x86-based Apple machine would *not* be the same as a PC.
PowerPC processors are much more powerful than x86 processors anyway.
There are very, very few benchmarks that support that assertion (and basically none if you move away from corner cases like Altivech enhanced photoshop filters and RC5 clients).
SPEC is the only popular benchmark that gives a reasonable cross-platform comparison for the general case (perversions like Apple’s recent testing methodologies aside). SPEC does not support your assertion. Compiler improvements may change this situation, but they haven’t yet.
Thats why the Navy bough a bunch of Xserves because they wanted them for the fast processing power that PowerPC processors could offer them.
Half true, although not in the sense you mean it. My guess would be they’ve got some of that corner-case hand-tuned software that derives enormouse benefits from Altivech (which would make sense given they’re supposed to be meant for signal/image processing) or they’ve got some other software that would be painful to port to x86 (less likely) – there’s few other compelling reasons to choose an Xserve over any number of name brand 1U x86 boxes or a simple embedded-processor black box for a custom application like that.
I’m sure it’s been discussed before and, yes, I’m about to discuss it again.
I do hope that someday in the future that Apple decides to release an x86 version of OS X.
Windows, in all it’s flavors, needs a serious alternative.
The only thing stopping me from getting a PowerMac G5 right now is the pricing.
Here in New Zealand I can get a fully topped out dual processor AthlonMP PC for $2900NZ including the monitor. Apple’s top of the line PowerMac G5 is $6,100 without a monitor!!!!!
From the benchmarks I’ve seen comparing the AthlonMP’s with the dual G5’s, the AthlonMP is faster *most* of the time.
I can’t see paying $3,000 more for *less* performance and hardware.
And please don’t talk to me about the stablity of XP compared to OS X. I run XP Pro here on both my systems. It’s fast, it’s stable and the only thing that knocks it flat are poorly written hardware drivers. Other then that, it’s rock solid.
Apple, I think you people make wonderful products, but the pricing is keeping the market share very low and leaving Microsoft/Intel right in the position they want to be in.
It’s time that we as computer users had another cost effective alternative to Microsoft.
If apple approached it like x86 world did…
you could run the g4 or ppc 970, period.
intel, amd?
cisc, risc?
It’s all about arch. Of course, then I’d certianly not mind clone builders- but you know apple has to do like everyone else did…
The best way is “OSX Ready”(or, OS XI) compliance specifications. Some machines naturally won’t run Mac OS, some might have the open firmwares spec, etc.. but whatever- Apple has a choice. Keep their arch “their ideal property, and pretty much limit market penetration”, or open it up to outsiders and act with them to mitigate and regulate specifications and compliance. And hey, if it’s like in the old days with isa clones not really being compatible, so what- you badmouth them when it’s crappy engineering, and you applaude when someone makes a “eureka!”