Wouldn't a custom chip with a translator from PPC to the Athlon "RISC" core work as well or better than the current Athlon? I still don't expect to see it, but I would think Apple could strike a deal to get this done if it needed to do it.
And this would be one reason Apple might work with AMD instead of Intel (someone questioned this logic in a thread, possibly this one).
I don't know if I'm really that worried about Apple falling behind. They aren't really doing that bad, and if they came out with good 1.4 ghz g4 chips in dual config, they would again be probably be up at the top:
I'm not really worried about them falling behind for my sake, but it will ultimately hurt business sales. If Maya or Photoshop or whatever ends up twice as fast in real terms (not just the MHZ) on another platform then there would probably be a more serious migration away from the platform. As it stands, we're treading water but with a bright future.
Its not really a choice. They have to implement that front end, to provide compatiblity with x86 apps.</strong><hr></blockquote>
Exactly. RISC and CISC are instruction architecture philosophies, not processor designs. A RISC processor can take advantage of the superiority inherent in a RISC ISA, and in the case of the Athlon the CISC program is essentially re-written to allow it to take advantage of some of the same features. But any machine that runs x86 code will always be a CISC machine. If the translator were moved off-chip then the Athlon would be considered a RISC processor, but as it now stands the Athlon can only be used in a CISC architecture and is therefore a CISC processor.
<strong>Sorry to sound condescending, but you apparenly know nothing about software development. LOL!!! Man, that gave me a good laugh!! </strong><hr></blockquote>
I'll admit I don't know much about Macintosh software development. But I've been developing Unix application software for 15 years. So I know something about software development in general, at least.
5 years ago I would have agreed with the idea that a CPU architecture switch was a major deal. But application software development has been steadily growing away from architecture-dependent code, i.e. assembly code, to object-oriented high-level code that should run on any architecture as long as you have the same underlying compiler design. There isn't a single OS X application that is more than a year old in design, so I would be shocked if there is any significant amount of architecture dependence in anything availble for OS X, other than the OS itself and device drivers. If I'm wrong, then most software developers are a lot further behind state-of-the-art than I would have believed.
And to change topics slightly, if I am wrong about how portable application code really is in general, then the switch from a 32-bit G4 to a 64-bit G5 architecture is going to be a major problem too. Although other postings here tell me that the G5 has a "32-bit compatible" mode to deal with that. But OS X would have to allow applications to use that mode, i.e. the kernel would have to support processes running in both 32-bit and 64-bit addressing simultaneously. Sounds a bit tricky to me. Would OS X on the G5 run _entirely_ in 32-bit mode to begin with?
BTW, I have already done some preliminary OS X (really, Darwin) ports of the applications I work on. They were originally designed for Intel/Linux. As far as the processor architecture is concerned, there was no issue at all, the ports were barely more than a re-compile (had to change a few header file references). Re-designing the interfaces is another matter altogether, but that has nothing to do with processor architecture and would not be an issue in a switch from PPC to AMD for existing OS X applications.
What if Apple made a server OS only on X86? No need for Classic nor Aqua. No need for developers to port consumer level apps. You *could* use a Classic-based GUI (in the sense of icons, behavior etc).
The lack of Classic and Aqua could serve as a deterrent for use by consumers. If necessary add some proprietary hardware lockin.
Is this technically possible? I think it would be the first sign of a move to X86. But I think X86 is a dead end anyway, but I know nothing. Nothing!
<strong>In all fairness, Darwin, Quicktime, OpenGL, and Java 2 all run on AMD processors right now. According to the graph, that's 3 out of the 5 lowest levels of the "taken care of". Quartz would be a big step but the audio probably would not.
I don't think this is going to happen, but I think it's an option Apple has prepared for in one way or another.</strong><hr></blockquote>
I'd wager that Apple has already done it. Remember the last "woe is us" period when G4s were stuck at a measly 400MHz for over a year? You think Apple didn't consider that they might have to abandon PowerPC, even back when they started OS X development? I'll bet that Quartz, and every bit of OS X for that matter, was done from the beginning with architecture-independence in mind.
Sure, Apple would like to stay with the PPC if possible. And if I'm wrong and Motorola can deliver significantly faster G4s, or even G5s, in the next few months, nobody will be happier than me. But since I don't believe that will happen, I don't think Apple has any choice but a processor switch, and I'm sure they are prepared to do it and could ship AMD-based Macs in a few months if they wanted to.
possible, however, remember that while software has become more and more portable over the years, drivers, which are still the heart of any computer, still require some heavy assembler coding and platform specific implementations.
JUst look at Windows XP. even on the same hardware, they have to rewrite the drivers, and it poses a major challenge to developers and users alike.
The best of all worlds, IBM, AMD and Motorola merge. Sony, Apple and Disney merge.
The nine states involved in the antitrust suit cry uncle, give up the law suit but pass a law that all future computer purchases can not include anything that runs Windows operating systems.
There is a joint resolution between all the above parties that Bill Gates must be taught a lesson and decide to make Apple the top OS and take the G5 to undescribable speeds.
It could happen.
about as believable as Apple porting Mac OS X to current intel and AMD chips.
<strong>...drivers, which are still the heart of any computer, still require some heavy assembler coding and platform specific implementations.</strong><hr></blockquote>
Agreed, and that would likely be the issue Apple is most concerned about if considering a CPU switch. The third-party hardware vendors have just gone through the pain of porting (more like re-writing from scratch) drivers from OS 9 to OS X. Now they have to do it all again... It should be much less of an effort, but it still takes resources.
On the other hand, I'm speculating that Apple does not abandon the PPC entirely. They continue to make iBooks and iMacs with G3 and G4 processors, and also a desktop tower machine at the low end of the line with the fastest G4 available. But at the middle and high end of the desktop line are machines with faster and fastest AMD (or sure, why not, Intel) processors. Maybe not all device drivers are available for those machines right away, but you can already do an awful lot with OS X just using the Apple-supplied built-in drivers. Which, for reasons I've already presented, I think Apple already has ported and ready to go.
<strong>The best of all worlds, IBM, AMD and Motorola merge. Sony, Apple and Disney merge.
The nine states involved in the antitrust suit cry uncle, give up the law suit but pass a law that all future computer purchases can not include anything that runs Windows operating systems.
There is a joint resolution between all the above parties that Bill Gates must be taught a lesson and decide to make Apple the top OS and take the G5 to undescribable speeds.
It could happen.
about as believable as Apple porting Mac OS X to current intel and AMD chips.
Heck, if your dreaming dream friggin BIG </strong><hr></blockquote>
Listen, Apple is not going to switch CPUs during the OS X transition. It would be chaos and it would piss off developers to no end. It's just not going to happen.
[quote] There is no G5. Not for at least a year, maybe never. The Apollo G4 might be coming, but it will only be a little faster than the current G4, not fast enough. Remember my disclaimer; I don't have anything more to go on here than rumor sites and news stories, plus common sense. The way I see it, Motorola took 2 years just to get the G4 from 400MHz to 800MHz. Now I am being asked to believe they will magically produce a 1.5-2GHz G5, the speed Apple needs for competitive Macs, a year or more sooner than expected. <hr></blockquote>
Yes, it took Apple 2 years to scale the G4 design from 400 to 800 MHz. But the G5 is a NEW design. With an entirely new design, it's possible to achieve such a jump in performance. There's nothing magic about it, the G5 will be fast by virtue of it's design features.
Motorola not only scaled the G4 during those 2 years, they learned about CPU design as well. R&D is not linear like you make it out to be. A design may get held up by a single problem, or a plethora of hitches, and when it's finished, it will appear to have come out of nowhere. But in reality moto has been working on the G5 for a while, and applying everything they learned from the G4 debacle.
Another thing to remember is that the incremental increases in CPU speed are driven as much by marketing as they are by engineering. Computer makers will make more money if they take little steps, rather than blowing their load as soon as it's available. However Apple's current predicament is an exception, I think, because they are loosing sales due to the GHz gap. So it would pay for them to blow their load and let us have it, so to speak.
I think Apple will be seeing CPU updates more frequently now. They're playing catch up. A switch is not likely. It would require too much work. Apple would lose its key reason why it has its chip, Altivec. Also, developers would have to write machine specific code to get their software to run on AMD. Imagine Apple saying, "Oh yes...now write Carbon apps for us. Wait...make them for AMD too." The only way a switch in suppliers would have been made would have been a year ago if Motorola didn't break the 500MHz hurdle. A CPU switch would be far more work than the conversion to OS X from classic OS's.
i don't understand the assumption that superbob and suckfuldotcom make about processor development. just because we're hearing about the apollo does not mean other chips aren't in development. do you actually believe that they work on one chip and work and work on it till it's "done" then decide to start working on the next chip?
motorola engineers:
"hmmm, let's just keep working on this g4 over here for another year. then once we're done with that, we'll get started on that g5 thing."
they've probably been working on it for a long time. in fact, i know they talked to apple years ago about using multi-core chips.
<strong>There isn't a single OS X application that is more than a year old in design, so I would be shocked if there is any significant amount of architecture dependence in anything availble for OS X, other than the OS itself and device drivers. If I'm wrong, then most software developers are a lot further behind state-of-the-art than I would have believed.</strong><hr></blockquote>
What about AltiVec? Quartz, Photoshop, FCP, and the iApps are heavily optimized for it, and I don't see Apple abandoning it for MHz (or GHz)
[quote]Yes, it took Apple 2 years to scale the G4 design from 400 to 800 MHz. But the G5 is a NEW design. With an entirely new design, it's possible to achieve such a jump in performance. There's nothing magic about it, the G5 will be fast by virtue of it's design features. <hr></blockquote>
Exactly. We don't know what's going on behind the scenes. For all we know, all of the G5 development has transfered to Apple and they already have a cheap, efficient foundry lined up. We shall see.
I, for one, am very excited about the future of the PPC. Not that I have the money to buy a G5.
"s it possible to replace the CISC instruction translator on the Athlon with a PowerPC RISC translator?"
I don't think the Power PC needs or uses any translators. It is native uniform sized instruction sets. That's not the problem.
There are plenty of other problems though. The compiler's registers are totally different.
The only way Apple could get processors from AMD or anybody else besides Motorola or IBM would be to be able to steal the current designs and Book E spec and ask them politely if they would consider investing millions in R&D and manufacture the ??? million chips/year.
Comments
And this would be one reason Apple might work with AMD instead of Intel (someone questioned this logic in a thread, possibly this one).
<a href="http://www.cpuscorecard.com/top_cpus.htm#IP4-2200"; target="_blank">http://www.cpuscorecard.com/top_cpus.htm#IP4-2200</a>;
<strong>Sorry to sound condescending, but you apparenly know nothing about software development. </strong><hr></blockquote>
There is nothing "condescending" about your observation. You must "talk down" to him to be condescending.
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
[edit: improved diction]
[ 01-10-2002: Message edited by: sjpsu ]</p>
<strong>
Its not really a choice. They have to implement that front end, to provide compatiblity with x86 apps.</strong><hr></blockquote>
Exactly. RISC and CISC are instruction architecture philosophies, not processor designs. A RISC processor can take advantage of the superiority inherent in a RISC ISA, and in the case of the Athlon the CISC program is essentially re-written to allow it to take advantage of some of the same features. But any machine that runs x86 code will always be a CISC machine. If the translator were moved off-chip then the Athlon would be considered a RISC processor, but as it now stands the Athlon can only be used in a CISC architecture and is therefore a CISC processor.
<strong>Sorry to sound condescending, but you apparenly know nothing about software development. LOL!!! Man, that gave me a good laugh!!
I'll admit I don't know much about Macintosh software development. But I've been developing Unix application software for 15 years. So I know something about software development in general, at least.
5 years ago I would have agreed with the idea that a CPU architecture switch was a major deal. But application software development has been steadily growing away from architecture-dependent code, i.e. assembly code, to object-oriented high-level code that should run on any architecture as long as you have the same underlying compiler design. There isn't a single OS X application that is more than a year old in design, so I would be shocked if there is any significant amount of architecture dependence in anything availble for OS X, other than the OS itself and device drivers. If I'm wrong, then most software developers are a lot further behind state-of-the-art than I would have believed.
And to change topics slightly, if I am wrong about how portable application code really is in general, then the switch from a 32-bit G4 to a 64-bit G5 architecture is going to be a major problem too. Although other postings here tell me that the G5 has a "32-bit compatible" mode to deal with that. But OS X would have to allow applications to use that mode, i.e. the kernel would have to support processes running in both 32-bit and 64-bit addressing simultaneously. Sounds a bit tricky to me. Would OS X on the G5 run _entirely_ in 32-bit mode to begin with?
BTW, I have already done some preliminary OS X (really, Darwin) ports of the applications I work on. They were originally designed for Intel/Linux. As far as the processor architecture is concerned, there was no issue at all, the ports were barely more than a re-compile (had to change a few header file references). Re-designing the interfaces is another matter altogether, but that has nothing to do with processor architecture and would not be an issue in a switch from PPC to AMD for existing OS X applications.
Bob
<strong>This guy sounds like a Wintel lover who is lusting after OS X but does not want to buy new hardware.</strong><hr></blockquote>
On my desk at home:
PowerMac G4/400, iBook G3/466, OS X. Soon to be added, a G4/800 iMac, unless something better comes out in the next 2 months.
On my desk at work:
PowerMac G4/450 Cube, OS X.
In my server closet at work:
PowerMac G3/466, OS X Server. Soon to be a dual G4/800, unless something better comes out in the next 2 months.
I Love Macs!
Bob
The lack of Classic and Aqua could serve as a deterrent for use by consumers. If necessary add some proprietary hardware lockin.
Is this technically possible? I think it would be the first sign of a move to X86. But I think X86 is a dead end anyway, but I know nothing. Nothing!
<strong>In all fairness, Darwin, Quicktime, OpenGL, and Java 2 all run on AMD processors right now. According to the graph, that's 3 out of the 5 lowest levels of the "taken care of". Quartz would be a big step but the audio probably would not.
I don't think this is going to happen, but I think it's an option Apple has prepared for in one way or another.</strong><hr></blockquote>
I'd wager that Apple has already done it. Remember the last "woe is us" period when G4s were stuck at a measly 400MHz for over a year? You think Apple didn't consider that they might have to abandon PowerPC, even back when they started OS X development? I'll bet that Quartz, and every bit of OS X for that matter, was done from the beginning with architecture-independence in mind.
Sure, Apple would like to stay with the PPC if possible. And if I'm wrong and Motorola can deliver significantly faster G4s, or even G5s, in the next few months, nobody will be happier than me. But since I don't believe that will happen, I don't think Apple has any choice but a processor switch, and I'm sure they are prepared to do it and could ship AMD-based Macs in a few months if they wanted to.
Bob
JUst look at Windows XP. even on the same hardware, they have to rewrite the drivers, and it poses a major challenge to developers and users alike.
G-News
The nine states involved in the antitrust suit cry uncle, give up the law suit but pass a law that all future computer purchases can not include anything that runs Windows operating systems.
There is a joint resolution between all the above parties that Bill Gates must be taught a lesson and decide to make Apple the top OS and take the G5 to undescribable speeds.
It could happen.
about as believable as Apple porting Mac OS X to current intel and AMD chips.
Heck, if your dreaming dream friggin BIG
<strong>...drivers, which are still the heart of any computer, still require some heavy assembler coding and platform specific implementations.</strong><hr></blockquote>
Agreed, and that would likely be the issue Apple is most concerned about if considering a CPU switch. The third-party hardware vendors have just gone through the pain of porting (more like re-writing from scratch) drivers from OS 9 to OS X. Now they have to do it all again... It should be much less of an effort, but it still takes resources.
On the other hand, I'm speculating that Apple does not abandon the PPC entirely. They continue to make iBooks and iMacs with G3 and G4 processors, and also a desktop tower machine at the low end of the line with the fastest G4 available. But at the middle and high end of the desktop line are machines with faster and fastest AMD (or sure, why not, Intel) processors. Maybe not all device drivers are available for those machines right away, but you can already do an awful lot with OS X just using the Apple-supplied built-in drivers. Which, for reasons I've already presented, I think Apple already has ported and ready to go.
Bob
<strong>The best of all worlds, IBM, AMD and Motorola merge. Sony, Apple and Disney merge.
The nine states involved in the antitrust suit cry uncle, give up the law suit but pass a law that all future computer purchases can not include anything that runs Windows operating systems.
There is a joint resolution between all the above parties that Bill Gates must be taught a lesson and decide to make Apple the top OS and take the G5 to undescribable speeds.
It could happen.
about as believable as Apple porting Mac OS X to current intel and AMD chips.
Heck, if your dreaming dream friggin BIG
There's a word for what you describe.
Utopia
[quote] There is no G5. Not for at least a year, maybe never. The Apollo G4 might be coming, but it will only be a little faster than the current G4, not fast enough. Remember my disclaimer; I don't have anything more to go on here than rumor sites and news stories, plus common sense. The way I see it, Motorola took 2 years just to get the G4 from 400MHz to 800MHz. Now I am being asked to believe they will magically produce a 1.5-2GHz G5, the speed Apple needs for competitive Macs, a year or more sooner than expected. <hr></blockquote>
Yes, it took Apple 2 years to scale the G4 design from 400 to 800 MHz. But the G5 is a NEW design. With an entirely new design, it's possible to achieve such a jump in performance. There's nothing magic about it, the G5 will be fast by virtue of it's design features.
Motorola not only scaled the G4 during those 2 years, they learned about CPU design as well. R&D is not linear like you make it out to be. A design may get held up by a single problem, or a plethora of hitches, and when it's finished, it will appear to have come out of nowhere. But in reality moto has been working on the G5 for a while, and applying everything they learned from the G4 debacle.
Another thing to remember is that the incremental increases in CPU speed are driven as much by marketing as they are by engineering. Computer makers will make more money if they take little steps, rather than blowing their load as soon as it's available. However Apple's current predicament is an exception, I think, because they are loosing sales due to the GHz gap. So it would pay for them to blow their load and let us have it, so to speak.
motorola engineers:
"hmmm, let's just keep working on this g4 over here for another year. then once we're done with that, we'll get started on that g5 thing."
they've probably been working on it for a long time. in fact, i know they talked to apple years ago about using multi-core chips.
Might there be a couple guys from Apple talking to AMD?
Wouldn't surprise me.
Is it possible to replace the CISC instruction translator on the Athlon with a PowerPC RISC translator?
Maybe
Will we see an AMD inside stickers on new Macs anytime soon?
No
<strong>There isn't a single OS X application that is more than a year old in design, so I would be shocked if there is any significant amount of architecture dependence in anything availble for OS X, other than the OS itself and device drivers. If I'm wrong, then most software developers are a lot further behind state-of-the-art than I would have believed.</strong><hr></blockquote>
What about AltiVec? Quartz, Photoshop, FCP, and the iApps are heavily optimized for it, and I don't see Apple abandoning it for MHz (or GHz)
[quote]Yes, it took Apple 2 years to scale the G4 design from 400 to 800 MHz. But the G5 is a NEW design. With an entirely new design, it's possible to achieve such a jump in performance. There's nothing magic about it, the G5 will be fast by virtue of it's design features. <hr></blockquote>
Exactly. We don't know what's going on behind the scenes. For all we know, all of the G5 development has transfered to Apple and they already have a cheap, efficient foundry lined up. We shall see.
I, for one, am very excited about the future of the PPC. Not that I have the money to buy a G5.
I don't think the Power PC needs or uses any translators. It is native uniform sized instruction sets. That's not the problem.
There are plenty of other problems though. The compiler's registers are totally different.
The only way Apple could get processors from AMD or anybody else besides Motorola or IBM would be to be able to steal the current designs and Book E spec and ask them politely if they would consider investing millions in R&D and manufacture the ??? million chips/year.
This might tick off IBM and Motorola though