Inspite of the fact Apple has said OS X will not run on anything but a Mac.
Actually that isn't quite the case.
Quote:
"We will not allow running Mac OS X on anything other than an Apple Mac." - Phil Schiller
This is VERY subtly different, it really only means Apple will not allow anyone to sell an OS X version ready to run on non-Apple x86, or someone could run an unallowed version of OS X.
Quote:
"Will I be able to run Windows/Linux/other x86 OS on my Intel Mac?" - Interviewer
"Most likely, but Apple won't support it. That doesn't preclude someone from running it on a Mac. They probably will. We won't do anything to preclude that." - Phil Schiller
Second quote pretty much rules out scorched earth motherboard lockdowns and VLIW CPU's. The comment in the context of expecting regular folks to run Itanium versions of Windows or Linux just doesn't make sense.
Why do I get so circumspect? Until the quantum security folks come up with a miracle, the basic laws of computer science show an automata that is implemented in hardware has an equivalent version that can be implemented in software. The hard part is deriving the equivalent version, but it exists. Any hardware tricks apple plays are expensive for them, and inevitably hackable. Why inevitably? Darwin. Apple supports Open Darwin which gets dedicated nerds most of the way there and removes 98%+ of the hard part.
Bottom line is Apple can't prevent OS X on non-Apple x86 forever, but they can manage the hacked demographic with success if they play their cards right.
You can be sure the Apple lawyers will fall all over anyone distributing a cobbled together pirate OS X binary. And there could very easily be techniques in the OS X Darwin layer that will cause the Open Darwin/OS X binary to break whenever you run Software Update, effectively preventing rampant piracy by non-hard-core hackers from continuing unabated. Sure hackers will keep working at it and counter-patching, but most script kiddies will get sick of not being able to use their box for a couple days every couple weeks and quit. The pirates that would use it for business, a HUGE problem in several parts of the world for MS, will get sick of lost productivity and go legit just to avoid the headaches, or just not use OS X any more. Any that go legit are more $$ lining Apples pockets, any that just quit weren't customers anyway.
But if you accept that 100% piracy avoidance is unattainable you can concentrate on managing how dedicated the pirate users are going to have to be over the medium to long run to continue using the pirate versions. All you have to do is make it too annoying for the lazy 80% and you have boutique viral marketing instead of widespread piracy.
And to tie it all back nicely into this thread. None of the above piracy management requires large outlays of effort, dollars, time or unpublished development tracks. And managed like this you don't have the issue of Apple vs. non-Apple machine cost, you want OS X you buy an Apple as the price of admission. It works today and there is no reason it can't continue to work that way, especially if Apple can further lower prices while maintaining their margins.
"We will not allow running Mac OS X on anything other than an Apple Mac." - Phil Schiller
This is VERY subtly different, it really only means Apple will not allow anyone to sell an OS X version ready to run on non-Apple x86, or someone could run an unallowed version of OS X.
I don't see the suptle difference. Schiller didn't say sell he said run, that is a clear difference.
Quote:
"Will I be able to run Windows/Linux/other x86 OS on my Intel Mac?" - Interviewer
"Most likely, but Apple won't support it. That doesn't preclude someone from running it on a Mac. They probably will. We won't do anything to preclude that." - Phil Schiller
Second quote pretty much rules out scorched earth motherboard lockdowns and VLIW CPU's. The comment in the context of expecting regular folks to run Itanium versions of Windows or Linux just doesn't make sense.
There is no desk top version of Windows or Linux for VLIW.
Intel has engineers in the lab perfecting Itanium's ability to decode x86 binaries in compiler software. So it would be possible to run Windows x86 and Linux x86 on Itanium.
If looking at the situation in a linear plane of logic Apple will use x86 processors.
Apple has its developers translating their software from Power to x86. So obviously that means the Macintels will run x86.
Which would be a straight forward logical assumption to make. And that may well be what Apple does.
But at the same time there are some holes and inconsistencies that don't make the x86 argument as obvious.
Such as the fact that Apple nor Intel have ever really said what chips will be used in Macintel's.
The fact that current developer tools do not address EM64T. It is doubtful Apple will develop 64 bit computing in PPC then reverse that development in its transition to Intel.
The fact that transitioning to x86 Apple will take on x86 legacy baggage.
The fact that developers machines are Pentium 4, its highly unlikely Macintel's will run P4's.
The fact that Intel wants to leave x86 and move to new instruction sets. Intel continues to develop Itanium inspite of the fact the industry at large mocks it as a failure.
The question should be asked. Why would Apple leave Power to move to an older inflexible architecture such as x86, when Intel even proclaims x86 does not have the future potential that Power will have?
Intel has positioned Itanium as the competitor to IBM's Power. Why would Apple not transition from Power to an architecture with equivalent potential future growth?
No one but Apple at this point can absolutely say.
They may use x86 processors, but I don't think any one can say that absolutely for sure.
My expectation is that Apple will do the unexpected. That has been Steve Jobs history.
"We will not allow running Mac OS X on anything other than an Apple Mac." - Phil Schiller
This is VERY subtly different, it really only means Apple will not allow anyone to sell an OS X version ready to run on non-Apple x86, or someone could run an unallowed version of OS X.
Quote:
Originally posted by TenoBell
I don't see the suptle difference. Schiller didn't say sell he said run, that is a clear difference.
I'd say "will not allow" is the subtly differentiating part of Schiller's quote, with emphasis on allow.
But I'd hesitate reading too much into that statement since what someone says at one time doesn't always accurately reflect what it'll actually mean at another (future) time. Language can be an awfully sloppy that way.
Sure Apple is telling the developers to compile to x86. They also say in the developer documentation on universal binaries that the ABI is "subject to change".
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW chip. All of the third-party developers' x86 apps will still run just fine, and they have the option of a single recompile to get a VLIW binary.
It won't be an Itanium; it will be a low-power chip that has the guts of an Itanium - namely VLIW architecture. Nobody except Apple wants the native VLIW feature, and Intel wants somebody to use it.
It's perfect for Apple because it will run Windows. Apple's hardware will run unix, Windows and OS X - nobody else will be able to do that. That's the key to moving users to OS X.
Sure Apple is telling the developers to compile to x86. They also say in the developer documentation on universal binaries that the ABI is "subject to change".
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW chip. All of the third-party developers' x86 apps will still run just fine, and they have the option of a single recompile to get a VLIW binary.
It won't be an Itanium; it will be a low-power chip that has the guts of an Itanium - namely VLIW architecture. Nobody except Apple wants the native VLIW feature, and Intel wants somebody to use it.
It's perfect for Apple because it will run Windows. Apple's hardware will run unix, Windows and OS X - nobody else will be able to do that. That's the key to moving users to OS X.
Could you please provide a road map of these chips. Intel publishes lots of stuff and I'm sure that they would like to have more than Apple as a customer, even if it is for embedded work, so I'm sure that they would advertise these chips. What is the road map of the first portable VLIW to the second and on and on. Where is the desktop and server road maps. If Intel wants to shift their whole production to these in the future they should be advertising them to no end. Maybe Intel would like to see more of a future so they bought a company that might help them. Intel buys lots of companies.
Sure Apple is telling the developers to compile to x86. They also say in the developer documentation on universal binaries that the ABI is "subject to change".
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW chip. All of the third-party developers' x86 apps will still run just fine, and they have the option of a single recompile to get a VLIW binary.
It won't be an Itanium; it will be a low-power chip that has the guts of an Itanium - namely VLIW architecture. Nobody except Apple wants the native VLIW feature, and Intel wants somebody to use it.
It's perfect for Apple because it will run Windows. Apple's hardware will run unix, Windows and OS X - nobody else will be able to do that. That's the key to moving users to OS X.
The OS X PPC ABI's were publicly listed as subject to change until 10.4 You will have to do much better than that.
Intel is squarely betting on the Next Generation architecture that runs x86 natively, not VLIW Itanium derivatives.
Intel is squarely betting on the Next Generation architecture that runs x86 natively, not VLIW Itanium derivatives.
Of course Intel has to cheerlead and rah rah its own product. I'd take press announcements with a grain of salt.
Somewhere there is a press announcement from years ago where Intel states Pentium 4 will hit 5Ghz.
Quote:
Could you please provide a road map of these chips. Intel publishes lots of stuff and I'm sure that they would like to have more than Apple as a customer, even if it is for embedded work, so I'm sure that they would advertise these chips.
Intel tried to present Itanium (VLIW) to everyone. But everyone rejected it. From what I've been reading Intel still desires to move away from x86, but is far less vocal about it now.
Apple is notoriously secretive. If Apple/Intel were working on an Itanium derivative, that revelation would be a huge shock to the industry. Intel may be willing to keep this a secret. The PC industry at large has scorned Itanium. Apple suddenly announcing a Mac running on a VLIW chip would be a great coup for Intel and its vindication of VLIW would be that much sweeter.
Lundy has a perfectly reasonable explanation of the benefits to Apple.
Quote:
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW. Apple's hardware will run Unix, Windows and OS X - nobody else will be able to do that.
Speaking for him, I think Lundy and I are only speculating. Much of this speculation is based on Apple's history and what seems to make sense for the future.
Apple are a very imaginative bunch and rarely do the obvious or expected. So I'm not looking at what is being presented and obvious. I'm looking for the wizard behind the curtain.
If Apple were just going to use Merom, Conroe, and Woodcrest, why continue to be so secretive about it?
Of course Apple may use x86. But at this point how can anyone say for sure Apple won't use something else?
About Intel "not advertising" it, their new microarchitecture says it has an advanced OOOE engine. And a lot less power consumption. Now the best way to do both of those things at the same time is VLIW with all of the OOOE in the compilation. Maybe the compiler is the "advanced OOOE engine".
Intel has announced a complete set of compilers just for Apple. I wonder why they would need to do that? Objective-C, yes.
Apple must know that a mobo chip isn't going to stop hackers. Every mobo chip has to be queried by the OS and it's just a matter of finding that query code in the OS and NOPing it.
About Intel "not advertising" it, their new microarchitecture says it has an advanced OOOE engine. And a lot less power consumption. Now the best way to do both of those things at the same time is VLIW with all of the OOOE in the compilation. Maybe the compiler is the "advanced OOOE engine".
So now we are down to the chips that Apple and others will be using are just VLIW chips cloaked as x86, so Yonah is an Itanium in costume.
Quote:
Intel has announced a complete set of compilers just for Apple. I wonder why they would need to do that? Objective-C, yes.
Just for Apple and for Windows and for Linux. Apparently Intel would like to make available compilers for people that use their chips.
Quote:
Apple must know that a mobo chip isn't going to stop hackers. Every mobo chip has to be queried by the OS and it's just a matter of finding that query code in the OS and NOPing it.
Adjustments could be made to keep this more secure. To think that Apple has asked Intel to make them a special chip so that they cannot be hacked, and that they covered their tracks by asking the developers to target x86, is more than speculation. There are a few too many twists and turns in your story. You believe that Apple seen a demonstration of VLIW and fell in love. I don't.
Brendon, how do you think the new microarchitecture is going to accomplish the energy savings? If not VLIW guts, what is the nature of it?
What I'm saying is that little by little Intel will attempt to try different things but their bread and butter is X86. For them to try unsucessful things is an invitation for AMD to take that bread and butter. Will they try different things like Itanium, yes. Will they bet the company on it, no. So they think that they have something that might, key word, might pan out to be something. So Intle will continue to invest in that technology, but their focus will continue to be X86. I have a hard time believing that they are willing to produce special chips only for Apple that they would like to sell to the greater market. No. Apple will use the Intel chips that we know about and they will use the Intel chipsets, and Intel will continue to evolve the hardware package, Apple will write the software that will take advantage of the HW. "It's all about the software." OSX is that good, what real options did Apple have? You make it sound like Intel had to have something really different to get Apple away from IBM. How long would it take for IBM to get a PPC to 3GHz and how long would it take them to get a mobile version on the market? PPC, did not pan out. If IBM were able to get a PPC to 3GHz and make a mobile version how much would it cost Apple for IBM to R&D those chips. Appel was paying IBM to R&D the PPC and Apple had to make their own chipset, if they wanted it to evolve. With Intel, Intel will foot the bill for R&D and will make the whole widget, chipset, MB, CPU, bluetooth, WiFi, WiMax, you name it. PPC was very expensive for Apple, to the point that they could not afford to continue to be there. I imagine that Sony and MS were huge distractions for IBM and that told Apple where IBM was really at. I can see that Apple was an easy switch to Intel, as others have said 'we will now be on equal footing', except for OSX. OSX is that good, what is HW without the OS? Get a copy of FreeBSD and run that for a while, see how productive you are. Intel will make the HW and Apple will make the great glue to bind it all together into a experience for the customer. Intel appears to be openingly looking at many new technologies like 4 cores on a die, but these may be mass produced. Intel is about mass production and the cost savings that brings. I think that Apple wanted to not have to worry about the CPU and finally get some savings instead of investing in IBM R&D, especially when IBM could not deliver. It was not some mystical new technology that got Apple it was the reality of where IBM was leading Apple. Does VLIW have a place, could be, let the market decide, apparently Intel thinks so because they are continueing to invest in it, but they are most likely wanting to improve Itanium and get Itanium into the market better. What you are suggesting is something that could, if it works, be something big years from now. X86 may not be able to go on for the next ten years, so yes Intel is trying to get something to take its place, but they are not going to try to become a chip boutique to do it. Intel is about mass production, when Intel has that stuff ready then maybe Apple will be there or maybe they will wait and see, I see evolution and the PentiumM is somethign that can evolve nicely with the lessons learned on the Pentium4. Clock speed was the battle cry through the last 5 to 7 years, now it is the PentiumM time and low power high output, that is the battle cry and only Intel will be able to pull this off they have the technology setting on shelf. The PentiumM will evolve and incorporate new technologies for the next 5 to 7 years and then there will be something new maybe Itanium and that chip family, maybe not time will tell there. For now it is hello Napa, which I'm loosely saying all things low power from Intel. That is what got Apple, the Intel road map of something that Apple liked, that Apple thinks will last for the next 5 to 7 years, maybe even 7 to 10 years. VLIW may be incorporated if it proves to be of benefit to X86 customers, until the opportune moment, this is not that opportune moment.
Intel has a new architecture (IA-64) it wants to push but few of its traditional customers will use.
Intel's partner Microsoft first agreed to support IA-64 only to turn and refuse to build an XP version of IA-64. Even worse Microsoft builds an XP - 64 version based on an ISA extension of Intel's rival (AMD64). Which forces Intel to support its rivals ISA extension.
Intel attempts to make a business deal with Apple Computer. Traditionally an enemy of Microsoft and Intel.
The advantage of Apple Computer is its independence from the larger traditional computer industry. This independence frees Apple to do things differently from all of Intel's traditional supply base.
The irony of Intel working with Apple. Is the fact that Apple would not be a large source of revenue for Intel. In fact Apple will only produce a small fraction of the revenue Intel gains from its largest customer (Dell). Which leaves the question: why does Intel want Apple as a partner?
Apple is independent of the Windows world. Apple has the freedom to compile its OS to any ISA and microchip architecture it chooses. But Apple's business in of itself will not make Intel much more profit.
Intel wins Apple Computer as a customer. Which direction should the Apple/Intel partnership go?
Intel could build a better VLIW Itanium based chip. That has multiprocessor, simultaneous multithreading, low power consumption, scales from server, desktop, to laptop. Decodes x86 binaries faster, which allows the chip to run all major desktop operating systems and software.
Because Apple Computer is known as a company that likes to shine and exhibit new technology Intel could exclusively offer such a chip for Apple to showcase its new architecture.
The other option is Intel could offer Apple Computer the same line of microprocessors it offers all of its supply base. In effect continue the industry's lock on x86 architecture. This effort would further support the 64 bit ISA of Intel's rival AMD. The results would do nothing to support Intel's ambition to push its own IA-64.
Apple Computer receiving its new standard x86 processors has diminished its traditional independence from the standard PC world. Apple would lock itself into the commodity hardware market and parallel hardware patterns with every other equipment manufacturer. Something Apple has never done in its history.
Quote:
What I'm saying is that little by little Intel will attempt to try different things but their bread and butter is X86.
This is true. To some degree it's Intel's fault. To a larger degree this is because of Microsoft.
Quote:
I have a hard time believing that they are willing to produce special chips only for Apple that they would like to sell to the greater market.
Itanium nor VLIW would be a special "Apple chip". Of course the long term idea for Intel is bring the architecture to a larger market. If and this is a big IF Apple were to use Itanium its purpose would be to introduce the architecture as viable to the entire PC market.
Quote:
You make it sound like Intel had to have something really different to get Apple away from IBM.
We certainly hope they did.
Quote:
How long would it take for IBM to get a PPC to 3GHz and how long would it take them to get a mobile version on the market? PPC, did not pan out.
Every company has finite time and finite resources.
We do know that Apple is not IBM's primary focus. IBM wanted to own the gaming console market and placed its resources into winning it. Of course if IBM had devoted resources into producing a 3Ghz G5 and a low power G5 for the PowerBook, it could have well done it.
Unfortunately for Apple the three major gaming consoles will yield IBM far more return than a 3Ghz G5 or a PowerBook G5.
Quote:
What you are suggesting is something that could, if it works, be something big years from now.
Quote:
Tukwila (Itanium 3) is the code-name for a future generation of Intel's Itanium processor family following Itanium 2 and Montecito. It is expected to come to market in 2007. Wikipedia
This is a new chip Intel is working on with engineers from the company it bought in Russia, the old Itanium team from HP, also a group of engineers who designed RISC chips (the same as Power PC) are working on this chip.
No this does not in itself prove this design is for Apple, but Apple would need a new design to run Itanium.
Quote:
VLIW may be incorporated if it proves to be of benefit to X86 customers, until the opportune moment, this is not that opportune moment.
You may be correct. Time will tell.
At this point Apple's options are open between x86 and VLIW. As it has not fully committed to either.
We have no idea what's going on in the background. It's possible Apple may give Intel time to get its new VLIW chips together.
If by 2006 Intel can't prove VLIW to be viable Apple can certainly go with x86.
You say VLIW is the only way. I say hello Yohna, because it is all about the software. OSX is very much ahead of Windows and that is only getting better, security is big stuff for Apple.
Perhaps you would should take your VLIW argument to Future Hardware since you are talking about hardware in the OSX forum. There is a good place to have this debate in future hardware, since what you are talking about is future hardware, here is a good thread -> Intel shows new chips, outlines platform directions
I've always maintained that I have no idea what Apple will do, neither does any one but Apple.
I've only said its a hypothesis and have laid out the logic to support that hypothesis. My support comes from what Apple and Intel both have stated they would like in the future.
The argument you've put forth is that Apple will use x86 and there is no way they will use VLIW. OS X encryption will inevitably be broken and Apple will have no choice but to sell OS X to the wide x86 market. Something Apple has said it has no desire to do.
Brendon, give it up. You are dealing with folks that have the same dogged determination as the Lone Gunmen.
They don't understand or acknowledge that cutting half the pipeline stages compared to Netburst can reduce the synchrionization circuitry by a factor of four or more, yielding huge power reductions. Power reductions not possible with the architecture they are touting. They don't believe Intel when it says it has a new architecture (not Itanium low-power) which supports the older ISA, and that that can possibly be a good thing.
There has to be this big industry secret, now that Steve is involved with Intel. The chipmaker is going to deliver misleading info or omit production plan roadmap entries altogether. Changing their 30+ year business philosophy of making sure the IT industry knows exactly what is coming, when, so industry can better plan their upgrade cycles. Apple needs to materially decieve its developer base on how to compile its software to complete the charade.
It obvious, the horse isn't dead! To them it's our illusion that there was a horse in the first place. It's mere coincidence the dog food and glue companies just held a mini-bidding war off to the side.
I had exactly the same thoughts and questions, lundy, after reading Hiro's "hostile" post.
Speaking of VLIW, I noticed it briefly mentioned in "The Inevitability of Xen" while reading a reprint of that paper in the August 2005 issue of Usenix ;login: yesterday afternoon. The first section was particularly interesting when considering it in the context of some discussion here about the possibilities for multiple OS support on Apple's future Intel-based products. If nothing else, it was enough to cast more doubt about that being used as a marketing strategy for increasing hardware sales/profits.
I'm curious if anyone who reads that paper (especially section 1) sees Xen[-like] "paravirtualization" somehow integrating with Apple's future hardware/software. Or maybe I'm just stretching too much to relate them, at least in this thread.
Comments
Originally posted by TenoBell
Inspite of the fact Apple has said OS X will not run on anything but a Mac.
Actually that isn't quite the case.
"We will not allow running Mac OS X on anything other than an Apple Mac." - Phil Schiller
This is VERY subtly different, it really only means Apple will not allow anyone to sell an OS X version ready to run on non-Apple x86, or someone could run an unallowed version of OS X.
"Will I be able to run Windows/Linux/other x86 OS on my Intel Mac?" - Interviewer
"Most likely, but Apple won't support it. That doesn't preclude someone from running it on a Mac. They probably will. We won't do anything to preclude that." - Phil Schiller
Second quote pretty much rules out scorched earth motherboard lockdowns and VLIW CPU's. The comment in the context of expecting regular folks to run Itanium versions of Windows or Linux just doesn't make sense.
Why do I get so circumspect? Until the quantum security folks come up with a miracle, the basic laws of computer science show an automata that is implemented in hardware has an equivalent version that can be implemented in software. The hard part is deriving the equivalent version, but it exists. Any hardware tricks apple plays are expensive for them, and inevitably hackable. Why inevitably? Darwin. Apple supports Open Darwin which gets dedicated nerds most of the way there and removes 98%+ of the hard part.
Bottom line is Apple can't prevent OS X on non-Apple x86 forever, but they can manage the hacked demographic with success if they play their cards right.
You can be sure the Apple lawyers will fall all over anyone distributing a cobbled together pirate OS X binary. And there could very easily be techniques in the OS X Darwin layer that will cause the Open Darwin/OS X binary to break whenever you run Software Update, effectively preventing rampant piracy by non-hard-core hackers from continuing unabated. Sure hackers will keep working at it and counter-patching, but most script kiddies will get sick of not being able to use their box for a couple days every couple weeks and quit. The pirates that would use it for business, a HUGE problem in several parts of the world for MS, will get sick of lost productivity and go legit just to avoid the headaches, or just not use OS X any more. Any that go legit are more $$ lining Apples pockets, any that just quit weren't customers anyway.
But if you accept that 100% piracy avoidance is unattainable you can concentrate on managing how dedicated the pirate users are going to have to be over the medium to long run to continue using the pirate versions. All you have to do is make it too annoying for the lazy 80% and you have boutique viral marketing instead of widespread piracy.
And to tie it all back nicely into this thread. None of the above piracy management requires large outlays of effort, dollars, time or unpublished development tracks. And managed like this you don't have the issue of Apple vs. non-Apple machine cost, you want OS X you buy an Apple as the price of admission. It works today and there is no reason it can't continue to work that way, especially if Apple can further lower prices while maintaining their margins.
"We will not allow running Mac OS X on anything other than an Apple Mac." - Phil Schiller
This is VERY subtly different, it really only means Apple will not allow anyone to sell an OS X version ready to run on non-Apple x86, or someone could run an unallowed version of OS X.
I don't see the suptle difference. Schiller didn't say sell he said run, that is a clear difference.
"Will I be able to run Windows/Linux/other x86 OS on my Intel Mac?" - Interviewer
"Most likely, but Apple won't support it. That doesn't preclude someone from running it on a Mac. They probably will. We won't do anything to preclude that." - Phil Schiller
Second quote pretty much rules out scorched earth motherboard lockdowns and VLIW CPU's. The comment in the context of expecting regular folks to run Itanium versions of Windows or Linux just doesn't make sense.
There is no desk top version of Windows or Linux for VLIW.
Intel has engineers in the lab perfecting Itanium's ability to decode x86 binaries in compiler software. So it would be possible to run Windows x86 and Linux x86 on Itanium.
Apple has its developers translating their software from Power to x86. So obviously that means the Macintels will run x86.
Which would be a straight forward logical assumption to make. And that may well be what Apple does.
But at the same time there are some holes and inconsistencies that don't make the x86 argument as obvious.
Such as the fact that Apple nor Intel have ever really said what chips will be used in Macintel's.
The fact that current developer tools do not address EM64T. It is doubtful Apple will develop 64 bit computing in PPC then reverse that development in its transition to Intel.
The fact that transitioning to x86 Apple will take on x86 legacy baggage.
The fact that developers machines are Pentium 4, its highly unlikely Macintel's will run P4's.
The fact that Intel wants to leave x86 and move to new instruction sets. Intel continues to develop Itanium inspite of the fact the industry at large mocks it as a failure.
The question should be asked. Why would Apple leave Power to move to an older inflexible architecture such as x86, when Intel even proclaims x86 does not have the future potential that Power will have?
Intel has positioned Itanium as the competitor to IBM's Power. Why would Apple not transition from Power to an architecture with equivalent potential future growth?
No one but Apple at this point can absolutely say.
They may use x86 processors, but I don't think any one can say that absolutely for sure.
My expectation is that Apple will do the unexpected. That has been Steve Jobs history.
"We will not allow running Mac OS X on anything other than an Apple Mac." - Phil Schiller
This is VERY subtly different, it really only means Apple will not allow anyone to sell an OS X version ready to run on non-Apple x86, or someone could run an unallowed version of OS X.
Originally posted by TenoBell
I don't see the suptle difference. Schiller didn't say sell he said run, that is a clear difference.
I'd say "will not allow" is the subtly differentiating part of Schiller's quote, with emphasis on allow.
But I'd hesitate reading too much into that statement since what someone says at one time doesn't always accurately reflect what it'll actually mean at another (future) time. Language can be an awfully sloppy that way.
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW chip. All of the third-party developers' x86 apps will still run just fine, and they have the option of a single recompile to get a VLIW binary.
It won't be an Itanium; it will be a low-power chip that has the guts of an Itanium - namely VLIW architecture. Nobody except Apple wants the native VLIW feature, and Intel wants somebody to use it.
It's perfect for Apple because it will run Windows. Apple's hardware will run unix, Windows and OS X - nobody else will be able to do that. That's the key to moving users to OS X.
Originally posted by lundy
Sure Apple is telling the developers to compile to x86. They also say in the developer documentation on universal binaries that the ABI is "subject to change".
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW chip. All of the third-party developers' x86 apps will still run just fine, and they have the option of a single recompile to get a VLIW binary.
It won't be an Itanium; it will be a low-power chip that has the guts of an Itanium - namely VLIW architecture. Nobody except Apple wants the native VLIW feature, and Intel wants somebody to use it.
It's perfect for Apple because it will run Windows. Apple's hardware will run unix, Windows and OS X - nobody else will be able to do that. That's the key to moving users to OS X.
Could you please provide a road map of these chips. Intel publishes lots of stuff and I'm sure that they would like to have more than Apple as a customer, even if it is for embedded work, so I'm sure that they would advertise these chips. What is the road map of the first portable VLIW to the second and on and on. Where is the desktop and server road maps. If Intel wants to shift their whole production to these in the future they should be advertising them to no end. Maybe Intel would like to see more of a future so they bought a company that might help them. Intel buys lots of companies.
Originally posted by lundy
Sure Apple is telling the developers to compile to x86. They also say in the developer documentation on universal binaries that the ABI is "subject to change".
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW chip. All of the third-party developers' x86 apps will still run just fine, and they have the option of a single recompile to get a VLIW binary.
It won't be an Itanium; it will be a low-power chip that has the guts of an Itanium - namely VLIW architecture. Nobody except Apple wants the native VLIW feature, and Intel wants somebody to use it.
It's perfect for Apple because it will run Windows. Apple's hardware will run unix, Windows and OS X - nobody else will be able to do that. That's the key to moving users to OS X.
The OS X PPC ABI's were publicly listed as subject to change until 10.4 You will have to do much better than that.
Intel is squarely betting on the Next Generation architecture that runs x86 natively, not VLIW Itanium derivatives.
Intel is squarely betting on the Next Generation architecture that runs x86 natively, not VLIW Itanium derivatives.
Of course Intel has to cheerlead and rah rah its own product. I'd take press announcements with a grain of salt.
Somewhere there is a press announcement from years ago where Intel states Pentium 4 will hit 5Ghz.
Could you please provide a road map of these chips. Intel publishes lots of stuff and I'm sure that they would like to have more than Apple as a customer, even if it is for embedded work, so I'm sure that they would advertise these chips.
Intel tried to present Itanium (VLIW) to everyone. But everyone rejected it. From what I've been reading Intel still desires to move away from x86, but is far less vocal about it now.
Apple is notoriously secretive. If Apple/Intel were working on an Itanium derivative, that revelation would be a huge shock to the industry. Intel may be willing to keep this a secret. The PC industry at large has scorned Itanium. Apple suddenly announcing a Mac running on a VLIW chip would be a great coup for Intel and its vindication of VLIW would be that much sweeter.
Lundy has a perfectly reasonable explanation of the benefits to Apple.
If Apple compiles OS X into VLIW binary, it can't be hacked, and it can't be run on any non-VLIW. Apple's hardware will run Unix, Windows and OS X - nobody else will be able to do that.
Speaking for him, I think Lundy and I are only speculating. Much of this speculation is based on Apple's history and what seems to make sense for the future.
Apple are a very imaginative bunch and rarely do the obvious or expected. So I'm not looking at what is being presented and obvious. I'm looking for the wizard behind the curtain.
If Apple were just going to use Merom, Conroe, and Woodcrest, why continue to be so secretive about it?
Of course Apple may use x86. But at this point how can anyone say for sure Apple won't use something else?
About Intel "not advertising" it, their new microarchitecture says it has an advanced OOOE engine. And a lot less power consumption. Now the best way to do both of those things at the same time is VLIW with all of the OOOE in the compilation. Maybe the compiler is the "advanced OOOE engine".
Intel has announced a complete set of compilers just for Apple. I wonder why they would need to do that? Objective-C, yes.
Apple must know that a mobo chip isn't going to stop hackers. Every mobo chip has to be queried by the OS and it's just a matter of finding that query code in the OS and NOPing it.
Originally posted by lundy
It's all a hypothesis.
About Intel "not advertising" it, their new microarchitecture says it has an advanced OOOE engine. And a lot less power consumption. Now the best way to do both of those things at the same time is VLIW with all of the OOOE in the compilation. Maybe the compiler is the "advanced OOOE engine".
So now we are down to the chips that Apple and others will be using are just VLIW chips cloaked as x86, so Yonah is an Itanium in costume.
Quote:
Intel has announced a complete set of compilers just for Apple. I wonder why they would need to do that? Objective-C, yes.
Just for Apple
Quote:
Apple must know that a mobo chip isn't going to stop hackers. Every mobo chip has to be queried by the OS and it's just a matter of finding that query code in the OS and NOPing it.
Adjustments could be made to keep this more secure. To think that Apple has asked Intel to make them a special chip so that they cannot be hacked, and that they covered their tracks by asking the developers to target x86, is more than speculation. There are a few too many twists and turns in your story. You believe that Apple seen a demonstration of VLIW and fell in love. I don't.
Originally posted by lundy
Brendon, how do you think the new microarchitecture is going to accomplish the energy savings? If not VLIW guts, what is the nature of it?
What I'm saying is that little by little Intel will attempt to try different things but their bread and butter is X86. For them to try unsucessful things is an invitation for AMD to take that bread and butter. Will they try different things like Itanium, yes. Will they bet the company on it, no. So they think that they have something that might, key word, might pan out to be something. So Intle will continue to invest in that technology, but their focus will continue to be X86. I have a hard time believing that they are willing to produce special chips only for Apple that they would like to sell to the greater market. No. Apple will use the Intel chips that we know about and they will use the Intel chipsets, and Intel will continue to evolve the hardware package, Apple will write the software that will take advantage of the HW. "It's all about the software." OSX is that good, what real options did Apple have? You make it sound like Intel had to have something really different to get Apple away from IBM. How long would it take for IBM to get a PPC to 3GHz and how long would it take them to get a mobile version on the market? PPC, did not pan out. If IBM were able to get a PPC to 3GHz and make a mobile version how much would it cost Apple for IBM to R&D those chips. Appel was paying IBM to R&D the PPC and Apple had to make their own chipset, if they wanted it to evolve. With Intel, Intel will foot the bill for R&D and will make the whole widget, chipset, MB, CPU, bluetooth, WiFi, WiMax, you name it. PPC was very expensive for Apple, to the point that they could not afford to continue to be there. I imagine that Sony and MS were huge distractions for IBM and that told Apple where IBM was really at. I can see that Apple was an easy switch to Intel, as others have said 'we will now be on equal footing', except for OSX. OSX is that good, what is HW without the OS? Get a copy of FreeBSD and run that for a while, see how productive you are. Intel will make the HW and Apple will make the great glue to bind it all together into a experience for the customer. Intel appears to be openingly looking at many new technologies like 4 cores on a die, but these may be mass produced. Intel is about mass production and the cost savings that brings. I think that Apple wanted to not have to worry about the CPU and finally get some savings instead of investing in IBM R&D, especially when IBM could not deliver. It was not some mystical new technology that got Apple it was the reality of where IBM was leading Apple. Does VLIW have a place, could be, let the market decide, apparently Intel thinks so because they are continueing to invest in it, but they are most likely wanting to improve Itanium and get Itanium into the market better. What you are suggesting is something that could, if it works, be something big years from now. X86 may not be able to go on for the next ten years, so yes Intel is trying to get something to take its place, but they are not going to try to become a chip boutique to do it. Intel is about mass production, when Intel has that stuff ready then maybe Apple will be there or maybe they will wait and see, I see evolution and the PentiumM is somethign that can evolve nicely with the lessons learned on the Pentium4. Clock speed was the battle cry through the last 5 to 7 years, now it is the PentiumM time and low power high output, that is the battle cry and only Intel will be able to pull this off they have the technology setting on shelf. The PentiumM will evolve and incorporate new technologies for the next 5 to 7 years and then there will be something new maybe Itanium and that chip family, maybe not time will tell there. For now it is hello Napa, which I'm loosely saying all things low power from Intel. That is what got Apple, the Intel road map of something that Apple liked, that Apple thinks will last for the next 5 to 7 years, maybe even 7 to 10 years. VLIW may be incorporated if it proves to be of benefit to X86 customers, until the opportune moment, this is not that opportune moment.
Intel has a new architecture (IA-64) it wants to push but few of its traditional customers will use.
Intel's partner Microsoft first agreed to support IA-64 only to turn and refuse to build an XP version of IA-64. Even worse Microsoft builds an XP - 64 version based on an ISA extension of Intel's rival (AMD64). Which forces Intel to support its rivals ISA extension.
Intel attempts to make a business deal with Apple Computer. Traditionally an enemy of Microsoft and Intel.
The advantage of Apple Computer is its independence from the larger traditional computer industry. This independence frees Apple to do things differently from all of Intel's traditional supply base.
The irony of Intel working with Apple. Is the fact that Apple would not be a large source of revenue for Intel. In fact Apple will only produce a small fraction of the revenue Intel gains from its largest customer (Dell). Which leaves the question: why does Intel want Apple as a partner?
Apple is independent of the Windows world. Apple has the freedom to compile its OS to any ISA and microchip architecture it chooses. But Apple's business in of itself will not make Intel much more profit.
Intel wins Apple Computer as a customer. Which direction should the Apple/Intel partnership go?
Intel could build a better VLIW Itanium based chip. That has multiprocessor, simultaneous multithreading, low power consumption, scales from server, desktop, to laptop. Decodes x86 binaries faster, which allows the chip to run all major desktop operating systems and software.
Because Apple Computer is known as a company that likes to shine and exhibit new technology Intel could exclusively offer such a chip for Apple to showcase its new architecture.
The other option is Intel could offer Apple Computer the same line of microprocessors it offers all of its supply base. In effect continue the industry's lock on x86 architecture. This effort would further support the 64 bit ISA of Intel's rival AMD. The results would do nothing to support Intel's ambition to push its own IA-64.
Apple Computer receiving its new standard x86 processors has diminished its traditional independence from the standard PC world. Apple would lock itself into the commodity hardware market and parallel hardware patterns with every other equipment manufacturer. Something Apple has never done in its history.
What I'm saying is that little by little Intel will attempt to try different things but their bread and butter is X86.
This is true. To some degree it's Intel's fault. To a larger degree this is because of Microsoft.
I have a hard time believing that they are willing to produce special chips only for Apple that they would like to sell to the greater market.
Itanium nor VLIW would be a special "Apple chip". Of course the long term idea for Intel is bring the architecture to a larger market. If and this is a big IF Apple were to use Itanium its purpose would be to introduce the architecture as viable to the entire PC market.
You make it sound like Intel had to have something really different to get Apple away from IBM.
We certainly hope they did.
How long would it take for IBM to get a PPC to 3GHz and how long would it take them to get a mobile version on the market? PPC, did not pan out.
Every company has finite time and finite resources.
We do know that Apple is not IBM's primary focus. IBM wanted to own the gaming console market and placed its resources into winning it. Of course if IBM had devoted resources into producing a 3Ghz G5 and a low power G5 for the PowerBook, it could have well done it.
Unfortunately for Apple the three major gaming consoles will yield IBM far more return than a 3Ghz G5 or a PowerBook G5.
What you are suggesting is something that could, if it works, be something big years from now.
Tukwila (Itanium 3) is the code-name for a future generation of Intel's Itanium processor family following Itanium 2 and Montecito. It is expected to come to market in 2007. Wikipedia
This is a new chip Intel is working on with engineers from the company it bought in Russia, the old Itanium team from HP, also a group of engineers who designed RISC chips (the same as Power PC) are working on this chip.
No this does not in itself prove this design is for Apple, but Apple would need a new design to run Itanium.
VLIW may be incorporated if it proves to be of benefit to X86 customers, until the opportune moment, this is not that opportune moment.
You may be correct. Time will tell.
At this point Apple's options are open between x86 and VLIW. As it has not fully committed to either.
We have no idea what's going on in the background. It's possible Apple may give Intel time to get its new VLIW chips together.
If by 2006 Intel can't prove VLIW to be viable Apple can certainly go with x86.
Originally posted by TenoBell
SNIP!
Horse == Dead
You say VLIW is the only way. I say hello Yohna, because it is all about the software. OSX is very much ahead of Windows and that is only getting better, security is big stuff for Apple.
Perhaps you would should take your VLIW argument to Future Hardware since you are talking about hardware in the OSX forum. There is a good place to have this debate in future hardware, since what you are talking about is future hardware, here is a good thread -> Intel shows new chips, outlines platform directions
I've always maintained that I have no idea what Apple will do, neither does any one but Apple.
I've only said its a hypothesis and have laid out the logic to support that hypothesis. My support comes from what Apple and Intel both have stated they would like in the future.
The argument you've put forth is that Apple will use x86 and there is no way they will use VLIW. OS X encryption will inevitably be broken and Apple will have no choice but to sell OS X to the wide x86 market. Something Apple has said it has no desire to do.
They don't understand or acknowledge that cutting half the pipeline stages compared to Netburst can reduce the synchrionization circuitry by a factor of four or more, yielding huge power reductions. Power reductions not possible with the architecture they are touting. They don't believe Intel when it says it has a new architecture (not Itanium low-power) which supports the older ISA, and that that can possibly be a good thing.
There has to be this big industry secret, now that Steve is involved with Intel. The chipmaker is going to deliver misleading info or omit production plan roadmap entries altogether. Changing their 30+ year business philosophy of making sure the IT industry knows exactly what is coming, when, so industry can better plan their upgrade cycles. Apple needs to materially decieve its developer base on how to compile its software to complete the charade.
It obvious, the horse isn't dead! To them it's our illusion that there was a horse in the first place. It's mere coincidence the dog food and glue companies just held a mini-bidding war off to the side.
I think he may be right.
Instead of bashing the intelligence or personality traits of the other posters, I would suggest that you spend your time explaining:
- why you don't think the new Intel architecture could be VLIW based
- how Apple can prevent running OS X on a vanilla PC if OS X is x86 binary
These are the issues being discussed, not the personal traits of the members.
Speaking of VLIW, I noticed it briefly mentioned in "The Inevitability of Xen" while reading a reprint of that paper in the August 2005 issue of Usenix ;login: yesterday afternoon. The first section was particularly interesting when considering it in the context of some discussion here about the possibilities for multiple OS support on Apple's future Intel-based products. If nothing else, it was enough to cast more doubt about that being used as a marketing strategy for increasing hardware sales/profits.
I'm curious if anyone who reads that paper (especially section 1) sees Xen[-like] "paravirtualization" somehow integrating with Apple's future hardware/software. Or maybe I'm just stretching too much to relate them, at least in this thread.