I agree that it's highly unlikely Apple will switch back, or offer its own line of PPC hardware as an alternative once Intel-based versions of all its hardware lines are on the market a couple of years down the road. However, I believe software support for PPC Macs will continue for years to come. Correct me if I'm wrong, but I don't believe Xcode in its current form even allows developers to create an Intel-only binary for Mac OS X; just a PPC binary or a PPC+Intel "fat" binary. So, in Xcode at least, maintaining support for legacy PPC hardware should be automatic.
I believe that the options are Intel, PPC, universal binary.
I'm quite sure that Apple and others will continue to keep universal binaries for several years, until not enough interest exists for it. But not a moment after. Developers, like all other business people look to cut their costs wherever, and whenever possible. If Apple came out with another PPC machine several years later (even two), they would rightfully feel betrayed. They have been told that Apple would have NO more PPC machines after 2007. And Apple had better mean it! They are looking to drop that extra work as soon as possible.
Customers also will not be happy. Now that Apple is trying to re-enter the enterprise market, they had better be consistent. No excuses.
This is more than a hobby machine. Hobbyists can fantasize. Business cannot!
I'm quite sure that Apple and others will continue to keep universal binaries for several years, until not enough interest exists for it. But not a moment after.
I agree here, with one exception. I don't believe that Apple will ever drop Mac OS X development for PowerPC. No, I don't mean market it and support it officially ad infinitum, but they will keep it in the lab, like they did with the Intel version. Perhaps some developers will do something similar.
I agree here, with one exception. I don't believe that Apple will ever drop Mac OS X development for PowerPC. No, I don't mean market it and support it officially ad infinitum, but they will keep it in the lab, like they did with the Intel version. Perhaps some developers will do something similar.
Well, I don't know but would it cost so much to have an experimental version for an alternative platform, like Apple did? And we are not talking about an OS here, which is a much more complex task.
It would cost significantly more to build chips for perhaps one quarter the number of machines. how much more would people pay for these machines? . . .
You present a good picture of engineering economics and the choices made in the design process. Thank you for sharing your knowledge and experience with us. You also make your case that Apple cannot afford a costly project for custom support chips when a product has a limited market. If it costs $5,000,000 to design and start producing a custom chip and 25,000 Macs are sold, there must be a $200 price increase on every Mac to paid back development cost. Obviously, not many products would be profitable with a burden like this.
Yet, as in many cost equations, if we change just one variable it can give totally different result. The case you present depends upon two conditions occurring simultaneously to be valid: first a significant development cost for a custom chip, and second low sales volume for the final product. Changing either of these variables makes a difference in the outcome. If there is no development cost, for example, there is no burden on the selling price and nothing needs to be paid back. Or, if the sales volume is 250,000 Mac, the custom chip represents only a $20 price increase on each Mac. In both of these cases, the equation indicates that it is now a very workable project.
It's clear that IBM would have the most influence on this equation through the design of their CPU. How many custom chips, if any, are needed? As IBM looks toward expanding the PPC into new markets, functions like on-chip memory controllers will be increasingly importance to potential customers. I believe that servers and high end workstations are two areas where Apple might consider building a future PPC Mac. Of course it depends on what kind of performance IBM can deliver. I certainly would not rule them out several years from now.
. . . [Developers] have been told that Apple would have NO more PPC machines after 2007. And Apple had better mean it! They are looking to drop that extra work as soon as possible. . .
Do you have a link for this statement? I don't remember ever hearing it or seeing it in print. As I understand it, Apple will transition all models of the Macs to Intel by 2007. I do not recall Apple saying that Macs will use Intel processors exclusively, however. It seems to me the door is open for two versions of some Macs, either Intel or PPC. Servers seem to be the most likely candidate, and possibly high end workstations.
.....It's clear that IBM would have the most influence on this equation through the design of their CPU. How many custom chips, if any, are needed? As IBM looks toward expanding the PPC into new markets, functions like on-chip memory controllers will be increasingly importance to potential customers. I believe that servers and high end workstations are two areas where Apple might consider building a future PPC Mac. Of course it depends on what kind of performance IBM can deliver. I certainly would not rule them out several years from now.
snoopy, i admire your confidence and continued faith in IBM, when some Mac users like me feel 'betrayed' and have 'given up' on them.
i think 2006 will be this critical time for IBM to see "what kind of performance IBM can deliver" and how it manages its chip business.
the way i see it, these are (rough ballpark) milestones that IBM has to hit for Apple to convince developers to hang on to PowerPC or even make it's push in the enterprise arena with post 970FX-PowerPC's.
July 2005-Dec 2005
-IBM provides clear direction and works well with apple to discuss g5 chips for Apple Macs in 2006
Jan 2006-June 2006
-Apple announces MacIntels, new g5 PPC products, of the g5 PPC products, IBM is able to meet and deliver on these with acceptable (to Apple) performance/watt, heat issues.
-IBM proposes initial Power6-based next gen PPC for possible Apple Macs. Steve says "I'll think about it"
-IBM preps for xbox360 and PS3
July 2006-Dec 2006
-We start to see first generation, mass-produced xbox360 and PS3 machines rolled out.
-IBM is able to ramp up production and deliver well on demand for these next-generation consoles.
-IBM continues to deliver to apple products that are in the pipeline. speedbumped 970fx's with same heat issues definitely no longer acceptable by middle of 2006.
-IBM demonstrates clear evidence and a convincing prototype of a Power6-based 9xxx that meets apple needs and budgets. Steve & gang approve such a thing
Jan 2007-June 2007
-Steve announces and celebrates full transition to MacIntel for almost all Mac products.
-Steve announces a enterprise-class series of servers, workstations, and maybe even terminals, based on Power6-derived PowerPC 9xxx architecture
-Such Mac Enterprise Edition products run faster, cooler, easier to maintain, backed by impressive IBM chip deliveries on mass-produced next-gen game consoles, and run Mac OS 10.5 for Enterprise superbly, kicks windows vista ass majorly.
Well, you'll excuse my dramatic flair and hyperbole, but essentially that's how i see things would have to pan out for PPC** to stay alive post-January 2007
Well, I don't know but would it cost so much to have an experimental version for an alternative platform, like Apple did? And we are not talking about an OS here, which is a much more complex task.
If a company is not building software for Linux, and there is VERY little commercial software out for it other than some highly specialized products such as warehouse control, finantial, etc., why would they have software development continue for a machine that doesn't even exist?
Usually both software and hardware companies wait to see what penetration something gets before they even propose a product. With tens of millions of machines out there, and at least 16 million Macs running OS X, why can't Apple convince more companies to make products for them? Why can't we even get prof 3D cards? That's only firmware and drivers. What could be simpler? Why won't banks and other services spend the 50 or so programmer hours to write software to allow us to use their services?
The answer is that they don't consider tens of millions of Mac's to be enough.
So now explain to me why they should write new programs for both x86 and PPC Macs, or even maintain them past the time they aren't selling enough to pay for their continued development.
For the almost zero chance that Apple may once again make a PPC machine?
You're forgetting that most Linux users prefere FOSS applications. I myself prefere to use GIMP instead of Adobe Photoshop, although it may not be better than Photoshop. The FOSS community has created, and sometimes cross-platformed, such amounts of software that it has become irrelevant wethere there is commercial software or not.
Take for example Nero: they have recently released NeroLINUX v2.0, but there doesn't seem to be much of a demand for it. K3b fills the spot nicely, as does GnomeBaker or Graveman. Or for example Adobe Reader (while a freeware, it is from a profit-based company), many users prefere either KPDF, or xPDF, or just Evince.
The problem is: Linux's model is different than Apple and the same argument does not apply. One could argue that Linux's desktop market share is negligible, but not so for Macs. 16 million computers are a lot of computers, and although some of them are going to be replaced by Intel-based Macs, the vast majority will still be PPC for at least 3 years to come. And guess what? The vast majority of them will want the new version of Office or Photoshop too.
You present a good picture of engineering economics and the choices made in the design process. Thank you for sharing your knowledge and experience with us. You also make your case that Apple cannot afford a costly project for custom support chips when a product has a limited market. If it costs $5,000,000 to design and start producing a custom chip and 25,000 Macs are sold, there must be a $200 price increase on every Mac to paid back development cost. Obviously, not many products would be profitable with a burden like this.
Yet, as in many cost equations, if we change just one variable it can give totally different result. The case you present depends upon two conditions occurring simultaneously to be valid: first a significant development cost for a custom chip, and second low sales volume for the final product. Changing either of these variables makes a difference in the outcome. If there is no development cost, for example, there is no burden on the selling price and nothing needs to be paid back. Or, if the sales volume is 250,000 Mac, the custom chip represents only a $20 price increase on each Mac. In both of these cases, the equation indicates that it is now a very workable project.
It's clear that IBM would have the most influence on this equation through the design of their CPU. How many custom chips, if any, are needed? As IBM looks toward expanding the PPC into new markets, functions like on-chip memory controllers will be increasingly importance to potential customers. I believe that servers and high end workstations are two areas where Apple might consider building a future PPC Mac. Of course it depends on what kind of performance IBM can deliver. I certainly would not rule them out several years from now.
These chips are expensive to make. There isn't just one chip that Apple would have to make, there are at least two. Plus Apple would have to continue to design their own mobo's. They would probably make a separate line of cases to distinguish the two lines. The PPC OS would most likely continue to use Open Firmware, while the x86 version most likely would use Intel's EFI. There could be other differences as well, who knows? If Apple thought that there was the slightest possibility that they might want to use another IBM chip, then they would have to continue development on All of their software. iLife, Studio, Shake, Filemaker, etc.
Again, you can't predict what IBM will do, and then use that as a reason to keep this going. Maybe even IBM doesn't know what they will do now. Despite the press, this was a blow to IBM's plans to spread the PPC far and wide as a hi end platform. Both IBM and Freescale are concentrating on embedded processors, as that's the main area of importance. A cpu to power Apple's machines is an aberration for them. With Apple buying 75% of those chips, there isn't much left. IBM was trying to gain more customers for those chips to help spread the cost. With Apple gone, it's a question as to whether it's worth even building them anymore. Even if IBM used them for their own servers, they would never sell enough to come close to making up for the loss of Apple.
So the question is not whether they will come out with some super PPC, but whether they will even continue making them after Apple leaves.
Remember all the press that said that Apple's leaving would make no financial difference to IBM? If that's so, and there is no reason not to believe it, then there would be NO reason for IBM to want Apple back. Even if they did continue making the chips. And Apple knows it.
Do you have a link for this statement? I don't remember ever hearing it or seeing it in print. As I understand it, Apple will transition all models of the Macs to Intel by 2007. I do not recall Apple saying that Macs will use Intel processors exclusively, however. It seems to me the door is open for two versions of some Macs, either Intel or PPC. Servers seem to be the most likely candidate, and possibly high end workstations.
You just repeated it. "Apple will transition ALL models".
There is no question there. That's what they will do.
If they transition all models, then by default that means they will not use PPC processors. Period.
snoopy, i admire your confidence and continued faith in IBM, when some Mac users like me feel 'betrayed' and have 'given up' on them.
i think 2006 will be this critical time for IBM to see "what kind of performance IBM can deliver" and how it manages its chip business.
the way i see it, these are (rough ballpark) milestones that IBM has to hit for Apple to convince developers to hang on to PowerPC or even make it's push in the enterprise arena with post 970FX-PowerPC's.
July 2005-Dec 2005
-IBM provides clear direction and works well with apple to discuss g5 chips for Apple Macs in 2006
Jan 2006-June 2006
-Apple announces MacIntels, new g5 PPC products, of the g5 PPC products, IBM is able to meet and deliver on these with acceptable (to Apple) performance/watt, heat issues.
-IBM proposes initial Power6-based next gen PPC for possible Apple Macs. Steve says "I'll think about it"
-IBM preps for xbox360 and PS3
July 2006-Dec 2006
-We start to see first generation, mass-produced xbox360 and PS3 machines rolled out.
-IBM is able to ramp up production and deliver well on demand for these next-generation consoles.
-IBM continues to deliver to apple products that are in the pipeline. speedbumped 970fx's with same heat issues definitely no longer acceptable by middle of 2006.
-IBM demonstrates clear evidence and a convincing prototype of a Power6-based 9xxx that meets apple needs and budgets. Steve & gang approve such a thing
Jan 2007-June 2007
-Steve announces and celebrates full transition to MacIntel for almost all Mac products.
-Steve announces a enterprise-class series of servers, workstations, and maybe even terminals, based on Power6-derived PowerPC 9xxx architecture
-Such Mac Enterprise Edition products run faster, cooler, easier to maintain, backed by impressive IBM chip deliveries on mass-produced next-gen game consoles, and run Mac OS 10.5 for Enterprise superbly, kicks windows vista ass majorly.
Well, you'll excuse my dramatic flair and hyperbole, but essentially that's how i see things would have to pan out for PPC** to stay alive post-January 2007
**edit: as in apple using PPC
Not you too, please.
The ONLY possibility that I can see is Some large academic or government customers asking them to keep the X Serves for a while longer. That's about it.
IBM won't come out with a super chip.
Jobs isn't about to say; " Oh, er, uh, IBM just put out this great new chip, and, well, I know I told you that they couldn't do it and all, but, well, it's just SO good that, uh, um, well, we're going to use it."
You're forgetting that most Linux users prefere FOSS applications. I myself prefere to use GIMP instead of Adobe Photoshop, although it may not be better than Photoshop. The FOSS community has created, and sometimes cross-platformed, such amounts of software that it has become irrelevant wethere there is commercial software or not.
Take for example Nero: they have recently released NeroLINUX v2.0, but there doesn't seem to be much of a demand for it. K3b fills the spot nicely, as does GnomeBaker or Graveman. Or for example Adobe Reader (while a freeware, it is from a profit-based company), many users prefere either KPDF, or xPDF, or just Evince.
The problem is: Linux's model is different than Apple and the same argument does not apply. One could argue that Linux's desktop market share is negligible, but not so for Macs. 16 million computers are a lot of computers, and although some of them are going to be replaced by Intel-based Macs, the vast majority will still be PPC for at least 3 years to come. And guess what? The vast majority of them will want the new version of Office or Photoshop too.
First of all, GIMP sucks. Maybe I should not prevaricate. GIMP SUCKS!!!
That's true for most Linux programs I've tried over the years. People who use them like to use them because they feel they are doing something organic (nothing personal, I just know a fair number of Linux users). The quality of the program is secondary, or tertiary.
Business users have options. They make choices based on matters that hobbiests don't consider. Why isn't Sun Office blazing its way? For that matter, why isn't Openoffice blazing ITS way? One reason is that they aren't very good. They look and feel like something out of Windows 3.1. And those are some of the better programs.
I would try a commercial PS job with you, if we were in the same room. I'll use CS 2, and you use GIMP. Guess what would happen?
Sure, I've been saying all along that most programs will be maintained until interest is down to the point of unprofitability. Or to the point where the company decides to forceably move their customers on to the new plarform. This will vary. It might take one year, or it might take four. But they will want to stop producing it.
[B]First of all, GIMP sucks. Maybe I should not prevaricate. GIMP SUCKS!!!
That is your opinion. For what I use it for, it's more than enough. It's not an image creation software, it's an image manipulation software. Hence, the M in GIMP. Gnu Image Manipulation Program.
Quote:
That's true for most Linux programs I've tried over the years.
Such as?
Quote:
People who use them like to use them because they feel they are doing something organic (nothing personal, I just know a fair number of Linux users). The quality of the program is secondary, or tertiary.
Talking out of your ass here. Provide factual evidence, not anecdotal.
Quote:
Business users have options. They make choices based on matters that hobbiests don't consider.
Linux is far from a hobby OS. BeOS is a hobby OS. Not Linux.
Quote:
Why isn't Sun Office blazing its way?
What does Sun have to do with Linux? We're not talking about Solaris here.
Quote:
For that matter, why isn't Openoffice blazing ITS way?
It takes a while to do even the simplest program out there, let alone a full-featured free Office Suite. Again, don't talk out of your ass. Show some basic understanding of program creation and advancement.
Quote:
One reason is that they aren't very good.
Again, that is your opinon. A lot of schools and government might disagree with you.
Quote:
They look and feel like something out of Windows 3.1.
This doesn't look like a Windows 3.1 program to me. Does it look like that to you? Or this. Or this for example.
Quote:
And those are some of the better programs.
You gonna name any of them, or should we use some weird mind-reading method to find out?
Quote:
I would try a commercial PS job with you, if we were in the same room. I'll use CS 2, and you use GIMP. Guess what would happen?
GIMP is not even competing with PS. They're two different beasts. And a lot depends on the skills of the artist as well.
Quote:
Sure, I've been saying all along that most programs will be maintained until interest is down to the point of unprofitability. Or to the point where the company decides to forceably move their customers on to the new plarform. This will vary. It might take one year, or it might take four. But they will want to stop producing it.
Big deal. In four years time, most people will want to upgrade their computers anyway. And buy MacIntels. Your point?
That is your opinion. For what I use it for, it's more than enough. It's not an image creation software, it's an image manipulation software. Hence, the M in GIMP. Gnu Image Manipulation Program.
Such as?
Talking out of your ass here. Provide factual evidence, not anecdotal.
Linux is far from a hobby OS. BeOS is a hobby OS. Not Linux.
What does Sun have to do with Linux? We're not talking about Solaris here.
It takes a while to do even the simplest program out there, let alone a full-featured free Office Suite. Again, don't talk out of your ass. Show some basic understanding of program creation and advancement.
Again, that is your opinon. A lot of schools and government might disagree with you.
This doesn't look like a Windows 3.1 program to me. Does it look like that to you? Or this. Or this for example.
You gonna name any of them, or should we use some weird mind-reading method to find out?
GIMP is not even competing with PS. They're two different beasts. And a lot depends on the skills of the artist as well.
Big deal. In four years time, most people will want to upgrade their computers anyway. And buy MacIntels. Your point?
Photoshop is an image manipulation program, it's just far more sophisticated than GIMP is. GIMP reminds me of programs I used on the WAX, better interface, but not much better tools. You compared it to Photoshop not me. If you said that all you used it for was basic work, I would have said fine.
Many Linux programs (or Open Source in general) that I've tried over the years have been clumsy, slow, missing major features, too complex etc.
A few in the image area that I remember because I tried them not TOO long ago are:
DevIL. It's so so, behind the times in features. Graphic Converter is much better all around.
iFactory works, but it crashes a fair amount and it's confusing at times, but it is beta.
Cthumb also works, but I would have thought better of it if it didn't come up with different results at times for the same collection of images, and it wasn't so slow on some. I can't find out the reasons. It's not consistent.
That's enough for now as it's late.
I love it when someone says "provide evidence" over what is, after all, an opinion, just like theirs is. I've gotten mine from talking to various Linux users over the years, as well as visiting Linux sites. One site OSNews, despite the name seems to be All Linux All The Time, no matter what the discussion is about. The arguments over vis and other editors can take pages. They talk about their asses a lot too, just as you do.
I didn't say that Linux was just a hobby OS. You said I said that. Business uses Linux for heavy duty lifting. Some programs are highly customized Open Source, but many are commercial. Hobbyists are different. Very few can modify the programs they use. And there are very few commercial programs for them, or for the desktop.
Sun has everything to do with Linux. OpenOffice which Linux users are touting as a replacement of Office is a direct descendant of Star Office from Sun. You did know that.
And stop being a jerk. If you want to converse then do so. It's different when one is calling a program a name. You don't have to insult me that way. your opinions are no better than mine. I had already said that it wasn't personal. What is your problem? This has nothing to do with you.
I know all about program development. OO has been out for years now. Most of what is is comes from Star Office. It wasn't written from scratch. It's typical Open Source. The infighting on the larger programs holds things back as well. There is a lot of ego there (in general, I mean).
And yes, again I will say that a lot of the programs aren't very good. Schools and the few governments that are using it are not using GIMP or programs like that. That's always been the problem with Linux, and why it's slow in moving out to the desktop in any big way. Very few commercial programs are available. Large users, as I've said can work around that.
Sure, some programs look fine. Most don't.
I already went through the program bit above. Maybe more later. Tomorrow.
Again, you did the comparison to Photoshop, not me. And yes, skill always counts.
I don't understand your last point at all. I'm saying that people are going to migrate to the x86 machines. The argument here has been that Apple will somehow come out with a new PPC machine several years after that.
I'm saying that they won't. That's what this entire software argument has been all about. Continued development. It had nothing to do with which is better. I only brought up Linux to make a point about commercial programs and numbers. The Linux users who write into the sites who aren't enamored with Open Source have been crying for commercial programs, drivers, etc.
. . . [Developers] have been told that Apple would have NO more PPC machines after 2007. And Apple had better mean it! . . .
We seem to have a problem with the English language here, so I quoted your statement again. I take this to mean you believe Apple promised developers it would never produce another PPC Mac after 2007. You believe Apple has permanently and irrevocably eliminated the PPC processor as a future option. (If so, it also means Steve has done an about face on his statement that he likes to have options.)
I see it differently. To me, Apple promised to transition all current Macs to Intel by 2007. For a time at least, all new Macs will have Intel processors. I see nothing in this that says new Macs will forever be Intel only. I don't believe Apple married Intel, but simply has a business contract, the terms of which we do not know. If further down the road the PPC processor outperforms Intel's top CPUs, why wouldn't Apple be interested?
The majority of Macs will remain PPC Macs for a number of years anyway. Apple could stick a new PPC motherboard into say an Xserve and sell it. Apple would not need to make a separate line of cases to distinguish the PPC and Intel lines, as you suggest. Both Xserves would run the same OS and software. Just as today there is no need to distinguish between a dual and single processor PowerMac, there would be no need to distinguish between a PPC and Intel Mac of any kind.
I see it differently. To me, Apple promised to transition all current Macs to Intel by 2007. For a time at least, all new Macs will have Intel processors. I see nothing in this that says new Macs will forever be Intel only. I don't believe Apple married Intel, but simply has a business contract, the terms of which we do not know. If further down the road the PPC processor outperforms Intel's top CPUs, why wouldn't Apple be interested?
Of course the Apple-Intel deal is not marriage but business, and there are terms etc. we don't know, and of course Apple is not going to use Intel chips for ever. But having Apple making a bold announcement about a transition for all its hardware lines and asking developers to migrate their software, would mean that, after having examined roadmaps and evaluated several other parameters, Apple has not the intention to come back to PowerPC for at least five years after the transition is complete. They must absolutely give enough time to developers so that their transition efforts pays off, before affording another transition or starting mixing architectures.
Otherwise it does not make sense and it is deadly dangerous. This is no joke. If Apple cannot take that seriously, then it is doomed. Not like so many times people said in the past, but really now.
. . . after having examined roadmaps and evaluated several other parameters, Apple has not the intention to come back to PowerPC for at least five years after the transition is complete. They must absolutely give enough time to developers so that their transition efforts pays off, before affording another transition or starting mixing architectures. . .
As I said above, "For a time at least, all new Macs will have Intel processors." We may just have different ideas about what a reasonable time is, five years or say two years. And there will never be and shouldn't be a transition back to only PPC Macs, IMHO. The advantages of Macs being on Intel are too overwhelming to even think about it. But that should not prevent Macs with mixed CPUs. Apple is making a point that a Mac is a Mac regardless of which CPU is under the hood. Following that line of logic, Apple should simply put the best CPU for the job under the hood. How many computer companies might wish to be in such a position? That is why I believe the PPC and Intel Macs will never look different. The only way to know is to look in the spec sheet.
As I said above, "For a time at least, all new Macs will have Intel processors." We may just have different ideas about what a reasonable time is, five years or say two years.
Exactly. For how many years Apple kept the PowerPC architecture? Around 10-12 years, despite the performance and yield issues that were not uncommon. Why expect something different now while going Intel? I would say that my estimate for 5 years after the completion of the transition, is rather conservative (that is it could be more).
Quote:
But that should not prevent Macs with mixed CPUs. Apple is making a point that a Mac is a Mac regardless of which CPU is under the hood. Following that line of logic, Apple should simply put the best CPU for the job under the hood. How many computer companies might wish to be in such a position? That is why I believe the PPC and Intel Macs will never look different. The only way to know is to look in the spec sheet.
The problem with mixed CPU architectures is not the look and feel of the machines. It is guaranteed that Apple will provide to the user a completely transparent system operation without needing to worry what is under the hood. The problem is that Apple is not the company that can afford to support more than one architecture, beyond the transition period. The other problem is the developers. Already asking them to switch is too much. Asking to support two architectures could lead to disaster. Or that's my opinion on the matter anyway.
Comments
Originally posted by Reid
I agree that it's highly unlikely Apple will switch back, or offer its own line of PPC hardware as an alternative once Intel-based versions of all its hardware lines are on the market a couple of years down the road. However, I believe software support for PPC Macs will continue for years to come. Correct me if I'm wrong, but I don't believe Xcode in its current form even allows developers to create an Intel-only binary for Mac OS X; just a PPC binary or a PPC+Intel "fat" binary. So, in Xcode at least, maintaining support for legacy PPC hardware should be automatic.
I believe that the options are Intel, PPC, universal binary.
I'm quite sure that Apple and others will continue to keep universal binaries for several years, until not enough interest exists for it. But not a moment after. Developers, like all other business people look to cut their costs wherever, and whenever possible. If Apple came out with another PPC machine several years later (even two), they would rightfully feel betrayed. They have been told that Apple would have NO more PPC machines after 2007. And Apple had better mean it! They are looking to drop that extra work as soon as possible.
Customers also will not be happy. Now that Apple is trying to re-enter the enterprise market, they had better be consistent. No excuses.
This is more than a hobby machine. Hobbyists can fantasize. Business cannot!
Originally posted by melgross
I'm quite sure that Apple and others will continue to keep universal binaries for several years, until not enough interest exists for it. But not a moment after.
I agree here, with one exception. I don't believe that Apple will ever drop Mac OS X development for PowerPC. No, I don't mean market it and support it officially ad infinitum, but they will keep it in the lab, like they did with the Intel version. Perhaps some developers will do something similar.
Originally posted by PB
I agree here, with one exception. I don't believe that Apple will ever drop Mac OS X development for PowerPC. No, I don't mean market it and support it officially ad infinitum, but they will keep it in the lab, like they did with the Intel version. Perhaps some developers will do something similar.
OS X, yes. Most everything else, no.
Originally posted by melgross
OS X, yes. Most everything else, no.
Well, I don't know but would it cost so much to have an experimental version for an alternative platform, like Apple did? And we are not talking about an OS here, which is a much more complex task.
Originally posted by Reid
So, in Xcode at least, maintaining support for legacy PPC hardware should be automatic.
Developers also have to have hardware to test on. No PPC hardware, no software or at best, untested PPC software.
Originally posted by melgross
It would cost significantly more to build chips for perhaps one quarter the number of machines. how much more would people pay for these machines? . . .
You present a good picture of engineering economics and the choices made in the design process. Thank you for sharing your knowledge and experience with us. You also make your case that Apple cannot afford a costly project for custom support chips when a product has a limited market. If it costs $5,000,000 to design and start producing a custom chip and 25,000 Macs are sold, there must be a $200 price increase on every Mac to paid back development cost. Obviously, not many products would be profitable with a burden like this.
Yet, as in many cost equations, if we change just one variable it can give totally different result. The case you present depends upon two conditions occurring simultaneously to be valid: first a significant development cost for a custom chip, and second low sales volume for the final product. Changing either of these variables makes a difference in the outcome. If there is no development cost, for example, there is no burden on the selling price and nothing needs to be paid back. Or, if the sales volume is 250,000 Mac, the custom chip represents only a $20 price increase on each Mac. In both of these cases, the equation indicates that it is now a very workable project.
It's clear that IBM would have the most influence on this equation through the design of their CPU. How many custom chips, if any, are needed? As IBM looks toward expanding the PPC into new markets, functions like on-chip memory controllers will be increasingly importance to potential customers. I believe that servers and high end workstations are two areas where Apple might consider building a future PPC Mac. Of course it depends on what kind of performance IBM can deliver. I certainly would not rule them out several years from now.
Originally posted by melgross
. . . [Developers] have been told that Apple would have NO more PPC machines after 2007. And Apple had better mean it! They are looking to drop that extra work as soon as possible. . .
Do you have a link for this statement? I don't remember ever hearing it or seeing it in print. As I understand it, Apple will transition all models of the Macs to Intel by 2007. I do not recall Apple saying that Macs will use Intel processors exclusively, however. It seems to me the door is open for two versions of some Macs, either Intel or PPC. Servers seem to be the most likely candidate, and possibly high end workstations.
Originally posted by snoopy
.....It's clear that IBM would have the most influence on this equation through the design of their CPU. How many custom chips, if any, are needed? As IBM looks toward expanding the PPC into new markets, functions like on-chip memory controllers will be increasingly importance to potential customers. I believe that servers and high end workstations are two areas where Apple might consider building a future PPC Mac. Of course it depends on what kind of performance IBM can deliver. I certainly would not rule them out several years from now.
snoopy, i admire your confidence and continued faith in IBM, when some Mac users like me feel 'betrayed' and have 'given up' on them.
i think 2006 will be this critical time for IBM to see "what kind of performance IBM can deliver" and how it manages its chip business.
the way i see it, these are (rough ballpark) milestones that IBM has to hit for Apple to convince developers to hang on to PowerPC or even make it's push in the enterprise arena with post 970FX-PowerPC's.
July 2005-Dec 2005
-IBM provides clear direction and works well with apple to discuss g5 chips for Apple Macs in 2006
Jan 2006-June 2006
-Apple announces MacIntels, new g5 PPC products, of the g5 PPC products, IBM is able to meet and deliver on these with acceptable (to Apple) performance/watt, heat issues.
-IBM proposes initial Power6-based next gen PPC for possible Apple Macs. Steve says "I'll think about it"
-IBM preps for xbox360 and PS3
July 2006-Dec 2006
-We start to see first generation, mass-produced xbox360 and PS3 machines rolled out.
-IBM is able to ramp up production and deliver well on demand for these next-generation consoles.
-IBM continues to deliver to apple products that are in the pipeline. speedbumped 970fx's with same heat issues definitely no longer acceptable by middle of 2006.
-IBM demonstrates clear evidence and a convincing prototype of a Power6-based 9xxx that meets apple needs and budgets. Steve & gang approve such a thing
Jan 2007-June 2007
-Steve announces and celebrates full transition to MacIntel for almost all Mac products.
-Steve announces a enterprise-class series of servers, workstations, and maybe even terminals, based on Power6-derived PowerPC 9xxx architecture
-Such Mac Enterprise Edition products run faster, cooler, easier to maintain, backed by impressive IBM chip deliveries on mass-produced next-gen game consoles, and run Mac OS 10.5 for Enterprise superbly, kicks windows vista ass majorly.
Well, you'll excuse my dramatic flair and hyperbole, but essentially that's how i see things would have to pan out for PPC** to stay alive post-January 2007
**edit: as in apple using PPC
Originally posted by PB
Well, I don't know but would it cost so much to have an experimental version for an alternative platform, like Apple did? And we are not talking about an OS here, which is a much more complex task.
If a company is not building software for Linux, and there is VERY little commercial software out for it other than some highly specialized products such as warehouse control, finantial, etc., why would they have software development continue for a machine that doesn't even exist?
Usually both software and hardware companies wait to see what penetration something gets before they even propose a product. With tens of millions of machines out there, and at least 16 million Macs running OS X, why can't Apple convince more companies to make products for them? Why can't we even get prof 3D cards? That's only firmware and drivers. What could be simpler? Why won't banks and other services spend the 50 or so programmer hours to write software to allow us to use their services?
The answer is that they don't consider tens of millions of Mac's to be enough.
So now explain to me why they should write new programs for both x86 and PPC Macs, or even maintain them past the time they aren't selling enough to pay for their continued development.
For the almost zero chance that Apple may once again make a PPC machine?
Take for example Nero: they have recently released NeroLINUX v2.0, but there doesn't seem to be much of a demand for it. K3b fills the spot nicely, as does GnomeBaker or Graveman. Or for example Adobe Reader (while a freeware, it is from a profit-based company), many users prefere either KPDF, or xPDF, or just Evince.
The problem is: Linux's model is different than Apple and the same argument does not apply. One could argue that Linux's desktop market share is negligible, but not so for Macs. 16 million computers are a lot of computers, and although some of them are going to be replaced by Intel-based Macs, the vast majority will still be PPC for at least 3 years to come. And guess what? The vast majority of them will want the new version of Office or Photoshop too.
Originally posted by snoopy
You present a good picture of engineering economics and the choices made in the design process. Thank you for sharing your knowledge and experience with us. You also make your case that Apple cannot afford a costly project for custom support chips when a product has a limited market. If it costs $5,000,000 to design and start producing a custom chip and 25,000 Macs are sold, there must be a $200 price increase on every Mac to paid back development cost. Obviously, not many products would be profitable with a burden like this.
Yet, as in many cost equations, if we change just one variable it can give totally different result. The case you present depends upon two conditions occurring simultaneously to be valid: first a significant development cost for a custom chip, and second low sales volume for the final product. Changing either of these variables makes a difference in the outcome. If there is no development cost, for example, there is no burden on the selling price and nothing needs to be paid back. Or, if the sales volume is 250,000 Mac, the custom chip represents only a $20 price increase on each Mac. In both of these cases, the equation indicates that it is now a very workable project.
It's clear that IBM would have the most influence on this equation through the design of their CPU. How many custom chips, if any, are needed? As IBM looks toward expanding the PPC into new markets, functions like on-chip memory controllers will be increasingly importance to potential customers. I believe that servers and high end workstations are two areas where Apple might consider building a future PPC Mac. Of course it depends on what kind of performance IBM can deliver. I certainly would not rule them out several years from now.
These chips are expensive to make. There isn't just one chip that Apple would have to make, there are at least two. Plus Apple would have to continue to design their own mobo's. They would probably make a separate line of cases to distinguish the two lines. The PPC OS would most likely continue to use Open Firmware, while the x86 version most likely would use Intel's EFI. There could be other differences as well, who knows? If Apple thought that there was the slightest possibility that they might want to use another IBM chip, then they would have to continue development on All of their software. iLife, Studio, Shake, Filemaker, etc.
Again, you can't predict what IBM will do, and then use that as a reason to keep this going. Maybe even IBM doesn't know what they will do now. Despite the press, this was a blow to IBM's plans to spread the PPC far and wide as a hi end platform. Both IBM and Freescale are concentrating on embedded processors, as that's the main area of importance. A cpu to power Apple's machines is an aberration for them. With Apple buying 75% of those chips, there isn't much left. IBM was trying to gain more customers for those chips to help spread the cost. With Apple gone, it's a question as to whether it's worth even building them anymore. Even if IBM used them for their own servers, they would never sell enough to come close to making up for the loss of Apple.
So the question is not whether they will come out with some super PPC, but whether they will even continue making them after Apple leaves.
Remember all the press that said that Apple's leaving would make no financial difference to IBM? If that's so, and there is no reason not to believe it, then there would be NO reason for IBM to want Apple back. Even if they did continue making the chips. And Apple knows it.
Originally posted by snoopy
Do you have a link for this statement? I don't remember ever hearing it or seeing it in print. As I understand it, Apple will transition all models of the Macs to Intel by 2007. I do not recall Apple saying that Macs will use Intel processors exclusively, however. It seems to me the door is open for two versions of some Macs, either Intel or PPC. Servers seem to be the most likely candidate, and possibly high end workstations.
You just repeated it. "Apple will transition ALL models".
There is no question there. That's what they will do.
If they transition all models, then by default that means they will not use PPC processors. Period.
You're playing now. All means all.
Let's stop this now.
Originally posted by sunilraman
snoopy, i admire your confidence and continued faith in IBM, when some Mac users like me feel 'betrayed' and have 'given up' on them.
i think 2006 will be this critical time for IBM to see "what kind of performance IBM can deliver" and how it manages its chip business.
the way i see it, these are (rough ballpark) milestones that IBM has to hit for Apple to convince developers to hang on to PowerPC or even make it's push in the enterprise arena with post 970FX-PowerPC's.
July 2005-Dec 2005
-IBM provides clear direction and works well with apple to discuss g5 chips for Apple Macs in 2006
Jan 2006-June 2006
-Apple announces MacIntels, new g5 PPC products, of the g5 PPC products, IBM is able to meet and deliver on these with acceptable (to Apple) performance/watt, heat issues.
-IBM proposes initial Power6-based next gen PPC for possible Apple Macs. Steve says "I'll think about it"
-IBM preps for xbox360 and PS3
July 2006-Dec 2006
-We start to see first generation, mass-produced xbox360 and PS3 machines rolled out.
-IBM is able to ramp up production and deliver well on demand for these next-generation consoles.
-IBM continues to deliver to apple products that are in the pipeline. speedbumped 970fx's with same heat issues definitely no longer acceptable by middle of 2006.
-IBM demonstrates clear evidence and a convincing prototype of a Power6-based 9xxx that meets apple needs and budgets. Steve & gang approve such a thing
Jan 2007-June 2007
-Steve announces and celebrates full transition to MacIntel for almost all Mac products.
-Steve announces a enterprise-class series of servers, workstations, and maybe even terminals, based on Power6-derived PowerPC 9xxx architecture
-Such Mac Enterprise Edition products run faster, cooler, easier to maintain, backed by impressive IBM chip deliveries on mass-produced next-gen game consoles, and run Mac OS 10.5 for Enterprise superbly, kicks windows vista ass majorly.
Well, you'll excuse my dramatic flair and hyperbole, but essentially that's how i see things would have to pan out for PPC** to stay alive post-January 2007
**edit: as in apple using PPC
Not you too, please.
The ONLY possibility that I can see is Some large academic or government customers asking them to keep the X Serves for a while longer. That's about it.
IBM won't come out with a super chip.
Jobs isn't about to say; " Oh, er, uh, IBM just put out this great new chip, and, well, I know I told you that they couldn't do it and all, but, well, it's just SO good that, uh, um, well, we're going to use it."
Oh yeah, baby!
Originally posted by Gene Clean
You're forgetting that most Linux users prefere FOSS applications. I myself prefere to use GIMP instead of Adobe Photoshop, although it may not be better than Photoshop. The FOSS community has created, and sometimes cross-platformed, such amounts of software that it has become irrelevant wethere there is commercial software or not.
Take for example Nero: they have recently released NeroLINUX v2.0, but there doesn't seem to be much of a demand for it. K3b fills the spot nicely, as does GnomeBaker or Graveman. Or for example Adobe Reader (while a freeware, it is from a profit-based company), many users prefere either KPDF, or xPDF, or just Evince.
The problem is: Linux's model is different than Apple and the same argument does not apply. One could argue that Linux's desktop market share is negligible, but not so for Macs. 16 million computers are a lot of computers, and although some of them are going to be replaced by Intel-based Macs, the vast majority will still be PPC for at least 3 years to come. And guess what? The vast majority of them will want the new version of Office or Photoshop too.
First of all, GIMP sucks. Maybe I should not prevaricate. GIMP SUCKS!!!
That's true for most Linux programs I've tried over the years. People who use them like to use them because they feel they are doing something organic (nothing personal, I just know a fair number of Linux users). The quality of the program is secondary, or tertiary.
Business users have options. They make choices based on matters that hobbiests don't consider. Why isn't Sun Office blazing its way? For that matter, why isn't Openoffice blazing ITS way? One reason is that they aren't very good. They look and feel like something out of Windows 3.1. And those are some of the better programs.
I would try a commercial PS job with you, if we were in the same room. I'll use CS 2, and you use GIMP. Guess what would happen?
Sure, I've been saying all along that most programs will be maintained until interest is down to the point of unprofitability. Or to the point where the company decides to forceably move their customers on to the new plarform. This will vary. It might take one year, or it might take four. But they will want to stop producing it.
Originally posted by melgross
[B]First of all, GIMP sucks. Maybe I should not prevaricate. GIMP SUCKS!!!
That is your opinion. For what I use it for, it's more than enough. It's not an image creation software, it's an image manipulation software. Hence, the M in GIMP. Gnu Image Manipulation Program.
That's true for most Linux programs I've tried over the years.
Such as?
People who use them like to use them because they feel they are doing something organic (nothing personal, I just know a fair number of Linux users). The quality of the program is secondary, or tertiary.
Talking out of your ass here. Provide factual evidence, not anecdotal.
Business users have options. They make choices based on matters that hobbiests don't consider.
Linux is far from a hobby OS. BeOS is a hobby OS. Not Linux.
Why isn't Sun Office blazing its way?
What does Sun have to do with Linux? We're not talking about Solaris here.
For that matter, why isn't Openoffice blazing ITS way?
It takes a while to do even the simplest program out there, let alone a full-featured free Office Suite. Again, don't talk out of your ass. Show some basic understanding of program creation and advancement.
One reason is that they aren't very good.
Again, that is your opinon. A lot of schools and government might disagree with you.
They look and feel like something out of Windows 3.1.
This doesn't look like a Windows 3.1 program to me. Does it look like that to you? Or this. Or this for example.
And those are some of the better programs.
You gonna name any of them, or should we use some weird mind-reading method to find out?
I would try a commercial PS job with you, if we were in the same room. I'll use CS 2, and you use GIMP. Guess what would happen?
GIMP is not even competing with PS. They're two different beasts. And a lot depends on the skills of the artist as well.
Sure, I've been saying all along that most programs will be maintained until interest is down to the point of unprofitability. Or to the point where the company decides to forceably move their customers on to the new plarform. This will vary. It might take one year, or it might take four. But they will want to stop producing it.
Big deal. In four years time, most people will want to upgrade their computers anyway. And buy MacIntels. Your point?
Originally posted by Gene Clean
That is your opinion. For what I use it for, it's more than enough. It's not an image creation software, it's an image manipulation software. Hence, the M in GIMP. Gnu Image Manipulation Program.
Such as?
Talking out of your ass here. Provide factual evidence, not anecdotal.
Linux is far from a hobby OS. BeOS is a hobby OS. Not Linux.
What does Sun have to do with Linux? We're not talking about Solaris here.
It takes a while to do even the simplest program out there, let alone a full-featured free Office Suite. Again, don't talk out of your ass. Show some basic understanding of program creation and advancement.
Again, that is your opinon. A lot of schools and government might disagree with you.
This doesn't look like a Windows 3.1 program to me. Does it look like that to you? Or this. Or this for example.
You gonna name any of them, or should we use some weird mind-reading method to find out?
GIMP is not even competing with PS. They're two different beasts. And a lot depends on the skills of the artist as well.
Big deal. In four years time, most people will want to upgrade their computers anyway. And buy MacIntels. Your point?
Photoshop is an image manipulation program, it's just far more sophisticated than GIMP is. GIMP reminds me of programs I used on the WAX, better interface, but not much better tools. You compared it to Photoshop not me. If you said that all you used it for was basic work, I would have said fine.
Many Linux programs (or Open Source in general) that I've tried over the years have been clumsy, slow, missing major features, too complex etc.
A few in the image area that I remember because I tried them not TOO long ago are:
DevIL. It's so so, behind the times in features. Graphic Converter is much better all around.
iFactory works, but it crashes a fair amount and it's confusing at times, but it is beta.
Cthumb also works, but I would have thought better of it if it didn't come up with different results at times for the same collection of images, and it wasn't so slow on some. I can't find out the reasons. It's not consistent.
That's enough for now as it's late.
I love it when someone says "provide evidence" over what is, after all, an opinion, just like theirs is. I've gotten mine from talking to various Linux users over the years, as well as visiting Linux sites. One site OSNews, despite the name seems to be All Linux All The Time, no matter what the discussion is about. The arguments over vis and other editors can take pages. They talk about their asses a lot too, just as you do.
I didn't say that Linux was just a hobby OS. You said I said that. Business uses Linux for heavy duty lifting. Some programs are highly customized Open Source, but many are commercial. Hobbyists are different. Very few can modify the programs they use. And there are very few commercial programs for them, or for the desktop.
Sun has everything to do with Linux. OpenOffice which Linux users are touting as a replacement of Office is a direct descendant of Star Office from Sun. You did know that.
And stop being a jerk. If you want to converse then do so. It's different when one is calling a program a name. You don't have to insult me that way. your opinions are no better than mine. I had already said that it wasn't personal. What is your problem? This has nothing to do with you.
I know all about program development. OO has been out for years now. Most of what is is comes from Star Office. It wasn't written from scratch. It's typical Open Source. The infighting on the larger programs holds things back as well. There is a lot of ego there (in general, I mean).
And yes, again I will say that a lot of the programs aren't very good. Schools and the few governments that are using it are not using GIMP or programs like that. That's always been the problem with Linux, and why it's slow in moving out to the desktop in any big way. Very few commercial programs are available. Large users, as I've said can work around that.
Sure, some programs look fine. Most don't.
I already went through the program bit above. Maybe more later. Tomorrow.
Again, you did the comparison to Photoshop, not me. And yes, skill always counts.
I don't understand your last point at all. I'm saying that people are going to migrate to the x86 machines. The argument here has been that Apple will somehow come out with a new PPC machine several years after that.
I'm saying that they won't. That's what this entire software argument has been all about. Continued development. It had nothing to do with which is better. I only brought up Linux to make a point about commercial programs and numbers. The Linux users who write into the sites who aren't enamored with Open Source have been crying for commercial programs, drivers, etc.
So, are you agreeing, or disagreeing?
Originally posted by melgross
. . . [Developers] have been told that Apple would have NO more PPC machines after 2007. And Apple had better mean it! . . .
We seem to have a problem with the English language here, so I quoted your statement again. I take this to mean you believe Apple promised developers it would never produce another PPC Mac after 2007. You believe Apple has permanently and irrevocably eliminated the PPC processor as a future option. (If so, it also means Steve has done an about face on his statement that he likes to have options.)
I see it differently. To me, Apple promised to transition all current Macs to Intel by 2007. For a time at least, all new Macs will have Intel processors. I see nothing in this that says new Macs will forever be Intel only. I don't believe Apple married Intel, but simply has a business contract, the terms of which we do not know. If further down the road the PPC processor outperforms Intel's top CPUs, why wouldn't Apple be interested?
The majority of Macs will remain PPC Macs for a number of years anyway. Apple could stick a new PPC motherboard into say an Xserve and sell it. Apple would not need to make a separate line of cases to distinguish the PPC and Intel lines, as you suggest. Both Xserves would run the same OS and software. Just as today there is no need to distinguish between a dual and single processor PowerMac, there would be no need to distinguish between a PPC and Intel Mac of any kind.
Originally posted by snoopy
I see it differently. To me, Apple promised to transition all current Macs to Intel by 2007. For a time at least, all new Macs will have Intel processors. I see nothing in this that says new Macs will forever be Intel only. I don't believe Apple married Intel, but simply has a business contract, the terms of which we do not know. If further down the road the PPC processor outperforms Intel's top CPUs, why wouldn't Apple be interested?
Of course the Apple-Intel deal is not marriage but business, and there are terms etc. we don't know, and of course Apple is not going to use Intel chips for ever. But having Apple making a bold announcement about a transition for all its hardware lines and asking developers to migrate their software, would mean that, after having examined roadmaps and evaluated several other parameters, Apple has not the intention to come back to PowerPC for at least five years after the transition is complete. They must absolutely give enough time to developers so that their transition efforts pays off, before affording another transition or starting mixing architectures.
Otherwise it does not make sense and it is deadly dangerous. This is no joke. If Apple cannot take that seriously, then it is doomed. Not like so many times people said in the past, but really now.
Originally posted by PB
. . . after having examined roadmaps and evaluated several other parameters, Apple has not the intention to come back to PowerPC for at least five years after the transition is complete. They must absolutely give enough time to developers so that their transition efforts pays off, before affording another transition or starting mixing architectures. . .
As I said above, "For a time at least, all new Macs will have Intel processors." We may just have different ideas about what a reasonable time is, five years or say two years. And there will never be and shouldn't be a transition back to only PPC Macs, IMHO. The advantages of Macs being on Intel are too overwhelming to even think about it. But that should not prevent Macs with mixed CPUs. Apple is making a point that a Mac is a Mac regardless of which CPU is under the hood. Following that line of logic, Apple should simply put the best CPU for the job under the hood. How many computer companies might wish to be in such a position? That is why I believe the PPC and Intel Macs will never look different. The only way to know is to look in the spec sheet.
Originally posted by snoopy
As I said above, "For a time at least, all new Macs will have Intel processors." We may just have different ideas about what a reasonable time is, five years or say two years.
Exactly. For how many years Apple kept the PowerPC architecture? Around 10-12 years, despite the performance and yield issues that were not uncommon. Why expect something different now while going Intel? I would say that my estimate for 5 years after the completion of the transition, is rather conservative (that is it could be more).
But that should not prevent Macs with mixed CPUs. Apple is making a point that a Mac is a Mac regardless of which CPU is under the hood. Following that line of logic, Apple should simply put the best CPU for the job under the hood. How many computer companies might wish to be in such a position? That is why I believe the PPC and Intel Macs will never look different. The only way to know is to look in the spec sheet.
The problem with mixed CPU architectures is not the look and feel of the machines. It is guaranteed that Apple will provide to the user a completely transparent system operation without needing to worry what is under the hood. The problem is that Apple is not the company that can afford to support more than one architecture, beyond the transition period. The other problem is the developers. Already asking them to switch is too much. Asking to support two architectures could lead to disaster. Or that's my opinion on the matter anyway.