PPC by AMD = interesting. I remember when AMD bought out a company by the name of NexGen which started the whole upward swing (it was a RISC processor that emulated CISC instructions, and quite fast too) toward the Athlon. Could be a possibility but I have a gut feeling that since IBM already makes G3s for Apple, that they will be the ones with the new chip for the new PowerMac. And to that end, I am guessing it will be a POWER 5 variant.</strong><hr></blockquote>
Rumour has it IBM has now got the rights to use Altivec, so with Book E architecture a custom CPU with the PPC instruction set is very possible.
As for Jobs comments, that was his way of leaving things open but in all honesty Apple going x86 is as likely as Texaco going solar!
<strong>runtime translation isn't necessary. translate ppc -> x86 and cache the resulting native (x86) binary, and it will run plently fast. </strong><hr></blockquote>
But not nearly as fast as code written and compiled for the x86 in the first place. It would put Apple right back in last place in terms of the speed race, and each processor upgrade would speed up the competition at least as much as it sped them up. No, this would be folly. It would also throw away the advantages that PPC does provide -- like lower power consumption & heat output, and AltiVec.
Steve meant exactly what he said: they want options, they like options. Options for choosing the processor and its supplier.
Given the decoder/core style of many of today's designs this seems like a more likely direction than moving the whole Mac user base to yet another instruction set architecture -- especially an old/creaking one. Imagine all the options they'd have if Moto, IBM, and AMD were all making PowerPCs. G4s from Moto, POWER5s & Sahara G3s from IBM, and new things from AMD (possibly an Operton w/ PPC decoder?). Apple invested a fair bit of money in Exponential back in the mid-90s in order to see what they could do with their new ideas for another PowerPC design... it didn't pan out then, but a third supplier can't be far from Apple's collective mind.
<strong>I wouldn't read too much into the quote, but in the immortal words of Arty Johnson it is "veerrrrryyyyy interesting".
</strong><hr></blockquote>
Amazingly no one has completed the quote and called the idea "..but alsho shtuhpid". I think it has some interesting parts to it myself.
[quote]<strong>
Also, what does Steve's quote"that first the company had to finish the transition to the OS X operating system, expected around the end of this year. " mean??????
What would require the "transition to the OS X operation system", before having options?????
</strong><hr></blockquote>
Anyone remember the rumor/Steveism *can't remember which it is* that a future evrsion of OS X will not allow OS 9 booting? Would this also take out Classic? If going beyond say 10.2.3 *achievable by December if the usual pace of a month or so per dot increase* would mean no more booting into 9 or even more drastically into Classic at all, then we would see MANY ticked off Mac users and developers.
Ok... this is definitely an amusing thread to read. I like to think about the future as much as anyone else, and this comment by SJ gives us all pause....
But for everyone saying "it can't be done" and "too much effort" etc. That's pretty much BS. Anyone who knows how OpenStep worked knows moving to a new CPU with OSX once EVERYTHING is OSX native - is a SNAP.
It was very, very, very common for an OpenStep app to be what they called a "fat binary." Like in the days with the transition from 68k to PPC - only this "fat binary" meant running on (usually) FOUR different CPUs - HP, 68k, Sparc, and x86. Cocoa's modular nature allowed this with very little overhead in terms of disk usage - and it just took a simple compile to get that stuff working on another platform. We're talking almost no work here.
So - after making the transition to OS X, it's true. Apple can have options - they can make a new platform on another chip, and everyone just has to build their app for 2 chips - EASY in the land of NeXT (nee OSX).
Oh, and as for the OS having stuff in it for i386, that's there for darwin, and some is in there because it was there in OpenStep 4.2 - it's sometimes better just to leave it alone than remove it and risk breaking some dependency that you haven't though of, especially when it doesn't take up much space, and that space isn't at a premium anymore anyway.
Just wondering how many of you who were planning on buying a new mac in the near future are still going to after reading this statement by Jobs? After reading this myself, I am now re-evaluating my plans to buy a new ppc Powermac, even if Apple does come out with an ass-kicking G5 next month. Even more serious is the affect this may have on the plans of the average computer user out there who was considering upgrading or buying a new mac. I'm sure many of them who have some computer knowledge are going to be wondering the same thing. This stupid statement by Jobs is absolutely not going to do anything to increase ppc mac sales for Apple.
Nice move Steve. Now you have many of us wondering if our ppc macs are going to be made obsolete by intel macs in the near future.
And what about the developers out there who are working or planning to make new ppc apps for OS X. Are you wondering if your time and investment in your new apps will going to be flushed down the drain if Apple moves to Intel?
Seems to me that Steve Jobs should retract this statement whether or not its true. Its an irresponsible statement and bad for business.
Personally I don't think SJ has the Brass-ones to do it.
But if its true. YEAH.
If Apple became a compiler writer this would be absolutely perfect.
Anyone could write anything and have it be multiplatform.
I do not think Apple would lose much in PPC sales because there is a fundamental quality in the PPC architecture that is respected by most people that work with it.
all it means to me is that apple will make faster computers in the future. This is already the case. This won't stop me a buying a new G4 XXX (?1.4 hopefully) with DDR if they come out in August. When new fatser machines come out I will evaluate them and determine if the speed/features warrant another upgrade. However, I use mine 99% for work and can evaluate based on how much time it takes to do certain operations. Increased speed is hard to evaluate in some cases beacaue it is a perception. However, when working with >500MB geotiff files speed is easy to evaluate.
My assumption and this is not coming from my friend or anyone else is this:
Consumer and Pro-Sumer Models: AMD type PPC processor with AltiVec and nVidia nForce. Don't forget that AMD and Mot are partners on many things and AMD and nVidia are very very close and Apple is very close to nVidia.
Servers: IBM Power4 or at least a variation of it.
That way Apple is still not putting all of thier eggs in one basket like they did with Mot. And that way Apple has a very very serious server processor.
To the person who said that Apple would not go with AMD cause they are not making money. Almost every single major corporation out there is having trouble right now including Motorola, Intel and AMD. (IBM posted a profit though ) So is Apple as well. That arguement means nothing.
Dual Boot. Man this has been the subject of many heated debates between my friend and I these past few days. He swears that at every turn with the people at Apple he knows and that he bullshits with about rumors, they do not deny this rumor, ever. One guy he spoke to recently just sat quiet for a miniute or two and then changed the subject, didn't even say "that's ridiculous" or anything like that. My friend said his reaction was more like: "how did that get out?"
Of course the next step would be a Virtual Machine server able to have each user pick what OS they want when they log in. All able to share resources simultaneously.
I won't hold my breath waiting, but it is certainly a feasible conclusion to where OS X could go in the future.
<strong>Ok... this is definitely an amusing thread to read. I like to think about the future as much as anyone else, and this comment by SJ gives us all pause....
But for everyone saying "it can't be done" and "too much effort" etc. That's pretty much BS. Anyone who knows how OpenStep worked knows moving to a new CPU with OSX once EVERYTHING is OSX native - is a SNAP.
It was very, very, very common for an OpenStep app to be what they called a "fat binary." Like in the days with the transition from 68k to PPC - only this "fat binary" meant running on (usually) FOUR different CPUs - HP, 68k, Sparc, and x86. Cocoa's modular nature allowed this with very little overhead in terms of disk usage - and it just took a simple compile to get that stuff working on another platform. We're talking almost no work here.
So - after making the transition to OS X, it's true. Apple can have options - they can make a new platform on another chip, and everyone just has to build their app for 2 chips - EASY in the land of NeXT (nee OSX).
Oh, and as for the OS having stuff in it for i386, that's there for darwin, and some is in there because it was there in OpenStep 4.2 - it's sometimes better just to leave it alone than remove it and risk breaking some dependency that you haven't though of, especially when it doesn't take up much space, and that space isn't at a premium anymore anyway.
While its true that NextStep/OpenStep went from the 68040 -> x86 -> PowerPC, that is a considerably different thing than from moving all of MacOSX and its applications over to a new platform (x86 or otherwise). People see some left over bits hanging around and think that "it wouldn't be that hard", but it would be an enormous undertaking and it would once again disrupt the Mac community and further fragment Apple's user base. It took Apple 4 years to get everything settled down after the PowerPC transition, and the debris from that is still hanging around. The Cocoa stuff might transition to a new platform reasonably well, but that leaves Carbon and a whole bunch of fairly new code which never existed on the OpenStep platform -- not to mention an ever increasing amount of AltiVec code which would need to be rewritten in assembly language to stay close to its current performance levels. 3rd parties would need to be on-board to transition their hardware drivers yet again.
Apple needs time between these transitions to focus on value-added features, and to let developers get on with the process of building better products. Just because they might be able to pull off such a transition as soon as next year, doesn't mean they should. I think Jobs' comment was simply saying: "okay, we've just about got the MacOSX situation under control now and then we'll be able to look around and evaluate what to do next". It isn't even close to meaning: "CONFIRMED, we're going to x86".
Dual boot is a wise transitional strategy, until Apple has a fast PPC option (Power 5, or a custom G5 design) or until it becomes clear that getting such a fast PPC option is not likely or not feasible. In the first case, they switch back to pure PPC, in the second they dump PPC and go with some other instruction set.
As for who would be providing the other instruction set... AMD/Nvidia is the most likely. It's even possible that a PPC version of Opteron is in the works, although the flexibility of OS X makes this not absolutely necessary - they could go with the standard Opteron plus Altivec.
Any app written in a high level language (C, C++, etc) for OS X, either Carbon or Cocoa, could be easily recomplied for x86 or the 64 bit extension of x86 used in Opteron. No rewrites necessary. The only sticking point would be PPC assembly... but I'm guessing the vast majority of apps that use PPC assembly are doing so to get at Altivec - so if Altivec units were present, those assembly instructions would still work, no code rewrite necessary. The only instructions that would be left out in the cold would be nonAltivec PPC assembly, and those could be translated - presumably these aren't very common, so most folks wouldn't notice a thing. Alternatively, if this is really an issue, just put a G3 in there to handle nonAltivec PPC assembly. But I really suspect the Opteron architecture could do a nice job with nonAltivec PPC instructions... a modest speed hit maybe, but not enough to really notice in the vast majority of applications. Besides, Opteron's raw speed might make up for the translation hit anyway (as in the 68000->PPC transition).
Anyone remember the rumor/Steveism *can't remember which it is* that a future evrsion of OS X will not allow OS 9 booting? Would this also take out Classic? If going beyond say 10.2.3 *achievable by December if the usual pace of a month or so per dot increase* would mean no more booting into 9 or even more drastically into Classic at all, then we would see MANY ticked off Mac users and developers.</strong><hr></blockquote>
You mean the 20+ million mac users who haven't upgraded to OSX yet will be upset that an OS they don't use won't boot 9 any longer?
<strong>Consumer and Pro-Sumer Models: AMD type PPC processor with AltiVec and nVidia nForce. Don't forget that AMD and Mot are partners on many things and AMD and nVidia are very very close and Apple is very close to nVidia.
</strong><hr></blockquote>
All sorts of arrangements could happen... perhaps AMD just becomes a fab for Moto so that Moto doesn't need to spend all the money to build these expensive fabs. They already did this with IBM. Perhaps they (AMD) will use this to slip into the PPC design business. If their core is general enough then they may want to maximize their return on investment in it, and spending a bit more (or having Apple fund) to put a PPC decoder on it could get them a new steady stream of revenue. Remember the chip margins in the x86 world have been cut to the bone, and beyond. AMD might be able to charge Apple a bit more since Apple actually has margins, unlike the PC makers.
[quote]<strong>
Servers: IBM Power4 or at least a variation of it.
</strong><hr></blockquote>
Just because the POWER4 is a server processor doesn't mean that IBM can't build really fast desktop processors. The core of the POWER4 is similar in concept and complexity to the PentiumIV. On the next generation of process (0.13 or 0.1 micron) we could see a POWER4-derived desktop PowerPC that can give Intel a run for its money. IBM hasn't built it yet because the machines they sell are servers. Given a customer who sells desktops, they would design a desktop chip.
[quote]<strong>
That way Apple is still not putting all of thier eggs in one basket like they did with Mot. And that way Apple has a very very serious server processor.
</strong><hr></blockquote>
In a way it still pigeonholes Apple though -- their servers are dependent on IBM and their desktops on somebody else, and their notebooks on somebody else. Better would be multiple choices for each line of product. If they can get all of their suppliers to build PowerPC w/ AltiVec & HyperTransport (for example) then they can suddenly mix and match parts depending on whatever their requirements are.
[quote]<strong>
To the person who said that Apple would not go with AMD cause they are not making money. Almost every single major corporation out there is having trouble right now including Motorola, Intel and AMD. (IBM posted a profit though ) So is Apple as well. That arguement means nothing.
</strong><hr></blockquote>
Except that AMD is probably hungry for more business!
[quote]<strong>
Dual Boot. Man this has been the subject of many heated debates between my friend and I these past few days. He swears that at every turn with the people at Apple he knows and that he bullshits with about rumors, they do not deny this rumor, ever. One guy he spoke to recently just sat quiet for a miniute or two and then changed the subject, didn't even say "that's ridiculous" or anything like that. My friend said his reaction was more like: "how did that get out?"</strong><hr></blockquote>
It could always be an internal Apple project that won't necessarily see the light of day. In a practical sense such a machine has never been a commercial success -- processors are fairly expensive (in terms of cost, space in the machine, power and heat) so any machine that has to have two of them is going to be at a disadvantage. Having two that use the same bus (i.e. SMP G4s) is much cheaper and more efficient. Running two OS's means you have to pay for two (in terms of money, memory, and disk space). OSX's fat binaries might let you have a mutant OS that knows how to load the appropriate binary onto two different processors, but that gets into all sorts of weird issues as well. Interesting project, but I really hope that Apple isn't seriously entertaining the idea -- if you think the Xserve is a "hack", its got nothing on this idea.
The Itanium isn't going to be an option, at least not for another generation or two -- except at the server level. When the EPIC project was in its early days didn't Jobs come out and say something about it? I only vaguely remember, but at one stage or another pretty much everybody was talking about it. Since then pretty much everybody has been ignoring it.
bah, powerpc is a nice platform if you have a nice implementation. this whole x86 humbug makes even less sense than a transition to macos 7 would. besides, it would probably be against Apples corporate identity.
"think different, use what everyone else does" yeah right
apple only needs a fat-bandwidth-bus attached G4 or better and those tasty Radeon 9700 cards.
Technically, it would be possible for Apple to go Intel.
Business wise, it would be a dumb idea. Here are some thoughts:
1) The only reason Apple would go Intel would be if Apple came to a realization that no PPC will ever be able to match or beat Intel for the forseeable future. While this is a possibility, I wouldn't bet on it.
2) Even suggesting this sort of thing could be similar to the MS nonsense where it publicly scolds Apple for not advertising OS X. In short, this could be a way of Jobs helping to light a fire under MOT's butt.
3) If Apple were to go Intel, they would be competing against every other commodity PC clone company where no innovation happens, just assemble boxes - cheaply! We all know Apple could not survive in this type of atmosphere. In short, Apple would become a software only company. With a much smaller budget, don't expect them to buy other companies and keep coming out with great new products and great updates to the OS nearly as often.
4) What about the developers? Apple has been pushing the OS X thing for some time. Apple would have to move OS X along quite a bit further (users adoption rate that is), before this would be possible.
5) Has nobody learned the lessons of OS/2? Great, OS X is better than Windows XP? So what, OS/2 was far better than Windows 3.1. The masses still standardized on Windows. Further, OS/2 lost because it could run Windows programs. This is why developers stopped making OS/2 specific versions of their software.
6) Lessons from Linux. This is a carry-over from the previous point. Yes, many will argue that Linux on the desktop is not ready for prime time. I'd even agree to that. However, more importantly, Linux boxes can be setup for dual boot. As long as it's possible for each machine to run Windows, the likelyhood of a Linux desktop version of any given software is just not very likely.
Anyway, the list goes one... For Apple to survive doing this, they would have to make their Intel based Macs not compatible with Windows, and hope that Windows doesn't make itself compatible with Intel based Apple machines. At the same time, users will scream out for better PC emulation, etc. Either way, it's a losing proposition for Apple.
No, instead, Apple would be better off designing their own PPC based chips and have them manufactured by someone else. Given the flexibility of MOT's Book E architecture, this just might be possible.
Comments
<strong>
PPC by AMD = interesting. I remember when AMD bought out a company by the name of NexGen which started the whole upward swing (it was a RISC processor that emulated CISC instructions, and quite fast too) toward the Athlon. Could be a possibility but I have a gut feeling that since IBM already makes G3s for Apple, that they will be the ones with the new chip for the new PowerMac. And to that end, I am guessing it will be a POWER 5 variant.</strong><hr></blockquote>
Rumour has it IBM has now got the rights to use Altivec, so with Book E architecture a custom CPU with the PPC instruction set is very possible.
As for Jobs comments, that was his way of leaving things open but in all honesty Apple going x86 is as likely as Texaco going solar!
<img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" /> <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
<strong>runtime translation isn't necessary. translate ppc -> x86 and cache the resulting native (x86) binary, and it will run plently fast. </strong><hr></blockquote>
But not nearly as fast as code written and compiled for the x86 in the first place. It would put Apple right back in last place in terms of the speed race, and each processor upgrade would speed up the competition at least as much as it sped them up. No, this would be folly. It would also throw away the advantages that PPC does provide -- like lower power consumption & heat output, and AltiVec.
Steve meant exactly what he said: they want options, they like options. Options for choosing the processor and its supplier.
Given the decoder/core style of many of today's designs this seems like a more likely direction than moving the whole Mac user base to yet another instruction set architecture -- especially an old/creaking one. Imagine all the options they'd have if Moto, IBM, and AMD were all making PowerPCs. G4s from Moto, POWER5s & Sahara G3s from IBM, and new things from AMD (possibly an Operton w/ PPC decoder?). Apple invested a fair bit of money in Exponential back in the mid-90s in order to see what they could do with their new ideas for another PowerPC design... it didn't pan out then, but a third supplier can't be far from Apple's collective mind.
<strong>I wouldn't read too much into the quote, but in the immortal words of Arty Johnson it is "veerrrrryyyyy interesting".
</strong><hr></blockquote>
Amazingly no one has completed the quote and called the idea "..but alsho shtuhpid". I think it has some interesting parts to it myself.
[quote]<strong>
Also, what does Steve's quote"that first the company had to finish the transition to the OS X operating system, expected around the end of this year. " mean??????
What would require the "transition to the OS X operation system", before having options?????
</strong><hr></blockquote>
Anyone remember the rumor/Steveism *can't remember which it is* that a future evrsion of OS X will not allow OS 9 booting? Would this also take out Classic? If going beyond say 10.2.3 *achievable by December if the usual pace of a month or so per dot increase* would mean no more booting into 9 or even more drastically into Classic at all, then we would see MANY ticked off Mac users and developers.
But for everyone saying "it can't be done" and "too much effort" etc. That's pretty much BS. Anyone who knows how OpenStep worked knows moving to a new CPU with OSX once EVERYTHING is OSX native - is a SNAP.
It was very, very, very common for an OpenStep app to be what they called a "fat binary." Like in the days with the transition from 68k to PPC - only this "fat binary" meant running on (usually) FOUR different CPUs - HP, 68k, Sparc, and x86. Cocoa's modular nature allowed this with very little overhead in terms of disk usage - and it just took a simple compile to get that stuff working on another platform. We're talking almost no work here.
So - after making the transition to OS X, it's true. Apple can have options - they can make a new platform on another chip, and everyone just has to build their app for 2 chips - EASY in the land of NeXT (nee OSX).
Oh, and as for the OS having stuff in it for i386, that's there for darwin, and some is in there because it was there in OpenStep 4.2 - it's sometimes better just to leave it alone than remove it and risk breaking some dependency that you haven't though of, especially when it doesn't take up much space, and that space isn't at a premium anymore anyway.
Well.... speculate on.....
THEY HAVE
The question is when will we have it ?
2003
2004
2late
Nice move Steve. Now you have many of us wondering if our ppc macs are going to be made obsolete by intel macs in the near future.
And what about the developers out there who are working or planning to make new ppc apps for OS X. Are you wondering if your time and investment in your new apps will going to be flushed down the drain if Apple moves to Intel?
Seems to me that Steve Jobs should retract this statement whether or not its true. Its an irresponsible statement and bad for business.
But if its true. YEAH.
If Apple became a compiler writer this would be absolutely perfect.
Anyone could write anything and have it be multiplatform.
I do not think Apple would lose much in PPC sales because there is a fundamental quality in the PPC architecture that is respected by most people that work with it.
Consumer and Pro-Sumer Models: AMD type PPC processor with AltiVec and nVidia nForce. Don't forget that AMD and Mot are partners on many things and AMD and nVidia are very very close and Apple is very close to nVidia.
Servers: IBM Power4 or at least a variation of it.
That way Apple is still not putting all of thier eggs in one basket like they did with Mot. And that way Apple has a very very serious server processor.
To the person who said that Apple would not go with AMD cause they are not making money. Almost every single major corporation out there is having trouble right now including Motorola, Intel and AMD. (IBM posted a profit though
Dual Boot. Man this has been the subject of many heated debates between my friend and I these past few days. He swears that at every turn with the people at Apple he knows and that he bullshits with about rumors, they do not deny this rumor, ever. One guy he spoke to recently just sat quiet for a miniute or two and then changed the subject, didn't even say "that's ridiculous" or anything like that. My friend said his reaction was more like: "how did that get out?"
LOL
I had actually forgotten about the rest of Arty Johnson's quote" "..but alsho shtuhpid""
I'm still laughing.
Thank you, you made my day <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />
<strong>...Dual Boot...</strong><hr></blockquote>
OS X makes that reality so such closer.
Of course the next step would be a Virtual Machine server able to have each user pick what OS they want when they log in. All able to share resources simultaneously.
I won't hold my breath waiting, but it is certainly a feasible conclusion to where OS X could go in the future.
<strong>Ok... this is definitely an amusing thread to read. I like to think about the future as much as anyone else, and this comment by SJ gives us all pause....
But for everyone saying "it can't be done" and "too much effort" etc. That's pretty much BS. Anyone who knows how OpenStep worked knows moving to a new CPU with OSX once EVERYTHING is OSX native - is a SNAP.
It was very, very, very common for an OpenStep app to be what they called a "fat binary." Like in the days with the transition from 68k to PPC - only this "fat binary" meant running on (usually) FOUR different CPUs - HP, 68k, Sparc, and x86. Cocoa's modular nature allowed this with very little overhead in terms of disk usage - and it just took a simple compile to get that stuff working on another platform. We're talking almost no work here.
So - after making the transition to OS X, it's true. Apple can have options - they can make a new platform on another chip, and everyone just has to build their app for 2 chips - EASY in the land of NeXT (nee OSX).
Oh, and as for the OS having stuff in it for i386, that's there for darwin, and some is in there because it was there in OpenStep 4.2 - it's sometimes better just to leave it alone than remove it and risk breaking some dependency that you haven't though of, especially when it doesn't take up much space, and that space isn't at a premium anymore anyway.
Well.... speculate on.....</strong><hr></blockquote>
While its true that NextStep/OpenStep went from the 68040 -> x86 -> PowerPC, that is a considerably different thing than from moving all of MacOSX and its applications over to a new platform (x86 or otherwise). People see some left over bits hanging around and think that "it wouldn't be that hard", but it would be an enormous undertaking and it would once again disrupt the Mac community and further fragment Apple's user base. It took Apple 4 years to get everything settled down after the PowerPC transition, and the debris from that is still hanging around. The Cocoa stuff might transition to a new platform reasonably well, but that leaves Carbon and a whole bunch of fairly new code which never existed on the OpenStep platform -- not to mention an ever increasing amount of AltiVec code which would need to be rewritten in assembly language to stay close to its current performance levels. 3rd parties would need to be on-board to transition their hardware drivers yet again.
Apple needs time between these transitions to focus on value-added features, and to let developers get on with the process of building better products. Just because they might be able to pull off such a transition as soon as next year, doesn't mean they should. I think Jobs' comment was simply saying: "okay, we've just about got the MacOSX situation under control now and then we'll be able to look around and evaluate what to do next". It isn't even close to meaning: "CONFIRMED, we're going to x86".
As for who would be providing the other instruction set... AMD/Nvidia is the most likely. It's even possible that a PPC version of Opteron is in the works, although the flexibility of OS X makes this not absolutely necessary - they could go with the standard Opteron plus Altivec.
Any app written in a high level language (C, C++, etc) for OS X, either Carbon or Cocoa, could be easily recomplied for x86 or the 64 bit extension of x86 used in Opteron. No rewrites necessary. The only sticking point would be PPC assembly... but I'm guessing the vast majority of apps that use PPC assembly are doing so to get at Altivec - so if Altivec units were present, those assembly instructions would still work, no code rewrite necessary. The only instructions that would be left out in the cold would be nonAltivec PPC assembly, and those could be translated - presumably these aren't very common, so most folks wouldn't notice a thing. Alternatively, if this is really an issue, just put a G3 in there to handle nonAltivec PPC assembly. But I really suspect the Opteron architecture could do a nice job with nonAltivec PPC instructions... a modest speed hit maybe, but not enough to really notice in the vast majority of applications. Besides, Opteron's raw speed might make up for the translation hit anyway (as in the 68000->PPC transition).
We live in interesting times...
<strong>
Anyone remember the rumor/Steveism *can't remember which it is* that a future evrsion of OS X will not allow OS 9 booting? Would this also take out Classic? If going beyond say 10.2.3 *achievable by December if the usual pace of a month or so per dot increase* would mean no more booting into 9 or even more drastically into Classic at all, then we would see MANY ticked off Mac users and developers.</strong><hr></blockquote>
You mean the 20+ million mac users who haven't upgraded to OSX yet will be upset that an OS they don't use won't boot 9 any longer?
<strong>Consumer and Pro-Sumer Models: AMD type PPC processor with AltiVec and nVidia nForce. Don't forget that AMD and Mot are partners on many things and AMD and nVidia are very very close and Apple is very close to nVidia.
</strong><hr></blockquote>
All sorts of arrangements could happen... perhaps AMD just becomes a fab for Moto so that Moto doesn't need to spend all the money to build these expensive fabs. They already did this with IBM. Perhaps they (AMD) will use this to slip into the PPC design business. If their core is general enough then they may want to maximize their return on investment in it, and spending a bit more (or having Apple fund) to put a PPC decoder on it could get them a new steady stream of revenue. Remember the chip margins in the x86 world have been cut to the bone, and beyond. AMD might be able to charge Apple a bit more since Apple actually has margins, unlike the PC makers.
[quote]<strong>
Servers: IBM Power4 or at least a variation of it.
</strong><hr></blockquote>
Just because the POWER4 is a server processor doesn't mean that IBM can't build really fast desktop processors. The core of the POWER4 is similar in concept and complexity to the PentiumIV. On the next generation of process (0.13 or 0.1 micron) we could see a POWER4-derived desktop PowerPC that can give Intel a run for its money. IBM hasn't built it yet because the machines they sell are servers. Given a customer who sells desktops, they would design a desktop chip.
[quote]<strong>
That way Apple is still not putting all of thier eggs in one basket like they did with Mot. And that way Apple has a very very serious server processor.
</strong><hr></blockquote>
In a way it still pigeonholes Apple though -- their servers are dependent on IBM and their desktops on somebody else, and their notebooks on somebody else. Better would be multiple choices for each line of product. If they can get all of their suppliers to build PowerPC w/ AltiVec & HyperTransport (for example) then they can suddenly mix and match parts depending on whatever their requirements are.
[quote]<strong>
To the person who said that Apple would not go with AMD cause they are not making money. Almost every single major corporation out there is having trouble right now including Motorola, Intel and AMD. (IBM posted a profit though
</strong><hr></blockquote>
Except that AMD is probably hungry for more business!
[quote]<strong>
Dual Boot. Man this has been the subject of many heated debates between my friend and I these past few days. He swears that at every turn with the people at Apple he knows and that he bullshits with about rumors, they do not deny this rumor, ever. One guy he spoke to recently just sat quiet for a miniute or two and then changed the subject, didn't even say "that's ridiculous" or anything like that. My friend said his reaction was more like: "how did that get out?"</strong><hr></blockquote>
It could always be an internal Apple project that won't necessarily see the light of day. In a practical sense such a machine has never been a commercial success -- processors are fairly expensive (in terms of cost, space in the machine, power and heat) so any machine that has to have two of them is going to be at a disadvantage. Having two that use the same bus (i.e. SMP G4s) is much cheaper and more efficient. Running two OS's means you have to pay for two (in terms of money, memory, and disk space). OSX's fat binaries might let you have a mutant OS that knows how to load the appropriate binary onto two different processors, but that gets into all sorts of weird issues as well. Interesting project, but I really hope that Apple isn't seriously entertaining the idea -- if you think the Xserve is a "hack", its got nothing on this idea.
The Itanium isn't going to be an option, at least not for another generation or two -- except at the server level. When the EPIC project was in its early days didn't Jobs come out and say something about it? I only vaguely remember, but at one stage or another pretty much everybody was talking about it. Since then pretty much everybody has been ignoring it.
"think different, use what everyone else does" yeah right
apple only needs a fat-bandwidth-bus attached G4 or better and those tasty Radeon 9700 cards.
Business wise, it would be a dumb idea. Here are some thoughts:
1) The only reason Apple would go Intel would be if Apple came to a realization that no PPC will ever be able to match or beat Intel for the forseeable future. While this is a possibility, I wouldn't bet on it.
2) Even suggesting this sort of thing could be similar to the MS nonsense where it publicly scolds Apple for not advertising OS X. In short, this could be a way of Jobs helping to light a fire under MOT's butt.
3) If Apple were to go Intel, they would be competing against every other commodity PC clone company where no innovation happens, just assemble boxes - cheaply! We all know Apple could not survive in this type of atmosphere. In short, Apple would become a software only company. With a much smaller budget, don't expect them to buy other companies and keep coming out with great new products and great updates to the OS nearly as often.
4) What about the developers? Apple has been pushing the OS X thing for some time. Apple would have to move OS X along quite a bit further (users adoption rate that is), before this would be possible.
5) Has nobody learned the lessons of OS/2? Great, OS X is better than Windows XP? So what, OS/2 was far better than Windows 3.1. The masses still standardized on Windows. Further, OS/2 lost because it could run Windows programs. This is why developers stopped making OS/2 specific versions of their software.
6) Lessons from Linux. This is a carry-over from the previous point. Yes, many will argue that Linux on the desktop is not ready for prime time. I'd even agree to that. However, more importantly, Linux boxes can be setup for dual boot. As long as it's possible for each machine to run Windows, the likelyhood of a Linux desktop version of any given software is just not very likely.
Anyway, the list goes one... For Apple to survive doing this, they would have to make their Intel based Macs not compatible with Windows, and hope that Windows doesn't make itself compatible with Intel based Apple machines. At the same time, users will scream out for better PC emulation, etc. Either way, it's a losing proposition for Apple.
No, instead, Apple would be better off designing their own PPC based chips and have them manufactured by someone else. Given the flexibility of MOT's Book E architecture, this just might be possible.
Steve
<img src="confused.gif" border="0">