Over the past 15 years the UNIX world has gone through this same conversion. I can't tell you the number of times I've heard all these same things said by the UNIX geeks as their favorite botique architectures have been replaced by x86 at the desktop, workstation, and small-to-mid sized server levels.
Well, except that all of those companies that transitioned are gone (DEC) or are mere shells of their former selves (SGI, Sun).
I agree with your overall sentiment, but I don't think that point is strong...
After milling this over two fitful sleepless nights (why can worrying about a f**ing computer do this to me), I am content and at peace with Apple's decision. It is the best. Are there risks? Of course.
The fact of the matter is that Apple could no longer be in the CPU/Bus Design business. It is FAR better to use the resources the rest of the world rely on. Altivec? Who cares. I used to, but after doing Altivec optimizations for my model (what a pain), I have realized that I would rather have a fast straight-up FPU. The G5 and AMD's have it now, the Intel chips coming out are rumored to surpass them both with ease. This leads to another point. Rumors regarding Intel chips tend to come true. The number of unfulfilled rumors, hope, dreaming, speculation from PPC the past 7 years have been frustrating with only a SINGLE breath of fresh air (G5).
The compilers, optimization libraries, etc. available on the x86 platform are incredible. Hell, we were all excited because Tiger was compiled with gcc 3.4 with more PPC optimization built-in. Building this support was hard work for Apple, now it is free.
Maybe, just maybe, Intel needs Apple. Intel has been at the mercy of Microsoft for too long now. Intel has always been Microsoft's biggest ally/enemy. The prepondance of cheap hardware with the ubiquity of Windows has built a great empire for them both; however, Microsoft is the more powerful as they are software only (excluding the xbox). The threats and manipulations that MS have placed on Intel over the years were exposed during the antitrust trial. Intel has and remains beholden to Microsoft's wishes. Intel thought that Netscape, Java, etc. might be their way out and they were put back into their place.
Intel has been pushing Linux, and will now presumably be an ally of MacOS X because they need software support from someone other than MS. I wouldn't be surprised if Intel gave Apple a big discount on the chips. Hell, they may have given Apple Dell like prices.
Well, except that all of those companies that transitioned are gone (DEC) or are mere shells of their former selves (SGI, Sun).
The market cap. of HP/Compaq is more than twice (almost 3 times) that of Apple.
Sun Microsystems is at least 1.5x of Apple.
Quote:
I agree with your overall sentiment, but I don't think that point is strong
I think that you missed my point. My point was not that the vendors of boutique MPU architectures were doing well in the market. My point was that the market had compelled the UNIX world to abandon those architectures and transition to x86.
The market cap. of HP/Compaq is more than twice (almost 3 times) that of Apple.
Sun Microsystems is at least 1.5x of Apple.
Well, to be annoying...
The market cap of Sun is 12B versus 30B for Apple. SGI is 214M. HP's market cap is 64B; however, just because Compaq bought DEC, I don't consider them the same company...
I did miss your point. You are exactly right, the market forced all of these companies (well, Sun still does SPARC development) to stop financing their own CPU's.
But probably not for much longer. I can imagine them going the Apple route (they already started down that path actually).
What makes all of this that much more interesting is the position it puts/keeps Intel in..."Monopolist to the Stars".
Of course both Apple and Sun are smart enough to keep the CPU hidden enough that switching around becomes much easier (Apple is probably doing this better than Sun). Either way, it seems that MS is in a much moer precarious position that way.
Intel would love to see Apple become more successful from a marketshare perspective. It means that there is a more diverse OS base running on their chip. Not just Windows. They must love Linux and Solaris too.
I think there could be some long term problems for MSFT.
Everyone is keeping options all over the place. Microsoft did this back before their OS dominance. One wonders if they are now (e.g., "secret double life for Office...running on Linux") or more concerned about protecting current business.
The market cap of Sun is 12B versus 30B for Apple.
Yup, sorry about that. I read today's volume number not today's market cap.
And to be fair, I should have said that the MPU architectures for these vendors had been forced out of the merchant mobile/desktop/workstation/small-to-mid range server space.
Alpha is end-of-life; no designs for any applications.
PA-RISC is end-of-life; no designs for any applications.
MIPS still exists in the embedded space but there are no new MPU designs for non-embedded applications.
SPARC has been abandoned by Sun. Fujitsu will continue to develop newer generation SPARC derivative MPUs. It simply became too expensive for Sun to push Sparc forward and to work on the next generation architecture.
The path to where we are is strewn with the empty husks of the 'better' architectures that have fallen under the x86 boot. Alpha, MIPS, PA-RISC, SPARC, good bye and thanks for all the fish.
Saddest (tech) sentence I've read this year.
Applies to BeOS, OS/2, Oberon, Plan 9, etc under the heel of Windows, too.
There is no logic or justice in the world. Let's hope OS X doesn't end up as a husk as well.
Applies to BeOS, OS/2, Oberon, Plan 9, etc under the heel of Windows, too.
There is no logic or justice in the world. Let's hope OS X doesn't end up as a husk as well.
don't be so sad, dude. we as geeks have been programmed in the 90s to think that RISC is da bomb and CISC is dead... but such things are not as important in the next 10 years. in the next 10 years, it will be squeezing as much as possible out of 90nm, 65, 45?nm to deliver the cost, power, lowheat, etc. consumers demand ~ this will also translate upwards into the server space to some degree...
in 10 years time RISC vs x86 or whatever will be irrelevant with quantum, molecular, biological, nanotech, optical or whatever to bring us on.
we KNOW mac os X has a great future ahead of it, at least 5 years, and people are getting tired of windows, they just don't have any options. 1-10 years out, Intel seems to offer apple what it needs, and yes, i agree with previous posters that Intel can get out from under microsoft's thumb. linux-on-intel, macosX-on-intel, should be alright... mac os X-on-powerPC, well, i am mourning it now, but it's got a few more years left in it, a very safe, stable bet for at least 1-2 years until the dust settles.
To quote Johan De Gelas ( of Aceshardware.com fame ), from the final section of that review:
Quote:
The server performance of the Apple platform is, however, catastrophic. When we asked Apple for a reaction, they told us that some database vendors, Sybase and Oracle, have found a way around the threading problems. We'll try Sybase later, but frankly, we are very sceptical. The whole "multi-threaded Mach microkernel trapped inside a monolithic FreeBSD cocoon with several threading wrappers and coarse-grained threading access to the kernel", with a "backwards compatibility" millstone around its neck sounds like a bad fusion recipe for performance.
Ok, how about the CELL? Is IBM so much incompetent? In your way of thinking, why Apple suddenly left the PowerPC "Boat"?
Cell's first implementation is designed for a game console and does not deliver stellar performance on non-vectorized code. Its performance is not terrible, but obviously Apple wants top-notch performance on typical code. This doesn't mean IBM is incompetent, it means they have other goals (i.e. selling many millions of chips in consumer units). That is a distraction from what Apple needs, and that is something I've been afraid of since their involvement in Cell & XBox360 first became known.
I had been hoping that a middle ground could be reached and a Cell variant suited to Apple would keep Apple on the PPC course, but I have my own set of predjudices.
Quote:
Do you think this company is unaware of it? Don't you think INTEL has more tools for instance to lock some toys with DRM'ized Chipsets?
You don't know what DRM support IBM might be providing in Cell.
Quote:
PS: Sorry Programmer, Personnely, I'm very upset with Apple's hardware's choice.
Maybe I'm wrong? who knows.....
Unfortunately IBM's main business goals pretty much forced them into it.
What we have to understand here is that Intel is not the dog uninventive company that some think it is. As Ottelini (spelling?) shower, for those who might not have known, Intel inventor the DRAM, and the microprocessor itself.
Assuming you're talking about 1T cells, then I think IBM would have something to say about that. If you want to talk about DRAM in general, then you definitely cannot credit that to intel. They can claim to have the first commercial product and, of course, they get credit for the 4004.
Quote:
They were one of the originators of Flash memory and are the biggest supplier of that as well.
Again, I'm assuming that you're not limiting the discussion to NOR flash in which case that honor goes to Samsung
Quote:
Over the years Intel made several contributions to cpu design only to find MS in opposition to those innovations because they wanted to do it in software. Their software.
Intel came out with several SIMD extensions that were never used, and were therefore dropped. The ones in place now are the ones that MS supports, or work without software support.
If Intel was willing to take the risks over the years of coming out with innovations that they didn't know MS would support, they might be interested in doing that again. But this time with the support of Apple.
If this happens, and it's a greater possibility than some of what I've been reading here and elsewhere, then MS would feel compelled to support them as well, or lose performance to an x86 Mac OS.
Intel might be very happy at this turn of events because it will allow the to regain control of all of their chip development.
Who knows, this could be one of the reasons they've been wooing Apple for so long.
Intel has clearly been able to support their ISA for far longer than many have predicted, and they've done it with some very clever hacks. While I'm sure that Intel will continue to add instructions and functionality to their chips, MS certainly won't be intimidated by having new functions available. The only real threat to MS would be if Intel were to abandon x86 in favor of a more modern and elegant ISA. This simply isn't going to happen for a very long time, so this isn't really a credible threat and does not unencumber Intel from that burden.
Frankly what suprises me most about Jobs' statements is the comment about performance/W. Intel has to burn a decent amount of power to simply handle the instruction decoding which was always a clear advantage to PPC. The only interpretation of this that makes sense to me is in the portable space where PPC obviously hasn't kept pace. The other possibility I see is that Apple is looking to make more home media devices and believes that Intel has some rabbits in that area, but given PPC's successes in the embedded market, I'm skeptical.
I agree with previous posters that this announcement and the timeline laid out by Apple will significantly hurt Mac sales. Actually I predict that there will be a spike in sales followed by a drawn out lull as certain groups load up on the hardware they need for the next 3-4 years, including spares, followed by the unwashed masses not wanting to invest in a lame duck platform.
Finally, has anyone else noticed the inconsistency between Rosetta and a very late introduction of the first Mactels? If Rosetta is indeed that great, then why wouldn't Apple be releasing an Intel box before the end of the year just to mitigate any sales losses and to seriously refresh their powerbooks? Or is binary translation really not quite the panacea Jobs makes it out to be (shocking to think that Stevie might be exaggerating I know!), and in order for this switch to work Apple desperately needs the developers to make native binaries. My hypothetical money's on the latter.
Finally, has anyone else noticed the inconsistency between Rosetta and a very late introduction of the first Mactels? If Rosetta is indeed that great, then why wouldn't Apple be releasing an Intel box before the end of the year just to mitigate any sales losses and to seriously refresh their powerbooks? Or is binary translation really not quite the panacea Jobs makes it out to be (shocking to think that Stevie might be exaggerating I know!), and in order for this switch to work Apple desperately needs the developers to make native binaries. My hypothetical money's on the latter.
you have to remember that apple cannot afford to put all its eggs in intel products right now, once they decide to go intel intel has to commit in a certain way to support apple.
this means that there is still a significant level of investment (r&d, marketing, etc) in the powerpc product pipeline. i feel if they released intel-macs right now (even if it were possible), that would mean a lot of investment in powerpc goes straight into the trash.
with the current scenario.
1. they can recoup their investments on powerpc line by clearing their pipeline of powerpc Macs over next 1-2 years
2. they have powerPC fallback while intel Macs are developed
3. apple and intel have 6-12 months from today to deliver on intel-macs, this has to be pretty much flawless or apple is seriously screwed compared to xp/longhorn/linux on intel
4. they have to give time for the developers to come to grips with this
5. rosetta is very important but i agree, they have decided this is not enough. by apple's standards, fat binaries MUST be employed as far as possible*
*i think the core reason why rosetta won't cut it for the most part is because comparative benchmarks of many applications running on similarly spec'ed intel hardware may show windows/linux versions of those applications with a clear advantage over mac os X-rosetta'ed apps.
To quote Johan De Gelas ( of Aceshardware.com fame ), from the final section of that review:
The server performance of the Apple platform is, however, catastrophic. When we asked Apple for a reaction, they told us that some database vendors, Sybase and Oracle, have found a way around the threading problems. We'll try Sybase later, but frankly, we are very sceptical. The whole "multi-threaded Mach microkernel trapped inside a monolithic FreeBSD cocoon with several threading wrappers and coarse-grained threading access to the kernel", with a "backwards compatibility" millstone around its neck sounds like a bad fusion recipe for performance.
Agreed. Even IBM realized that mach was a waste of time and energy. Apple lost me, but not my wife, with the switch to X. I was never a fan of the nExt architecture and by extension X, and the few interface gains in X were offset by the losses. X could and still can't make its mind up. Is it a unix box? Is it mach? All of that has limited what could be done, and now that the hardware will be identical does open Apple up to some potentially unflattering comparisons. For all the talk in this thread about giving credit to Intel for producing competitive hardware, everyone should be willing to admit that MS can produce very competitive performance, as the linux folks have seen from time to time, especially in the enterprise space.
Agreed. Even IBM realized that mach was a waste of time and energy...
Yup, me and Mach both left CMU at about the same time, so I had a bit of a soft spot for it. But facts-is-facts.
Quote:
For all the talk in this thread about giving credit to Intel for producing competitive hardware, everyone should be willing to admit that MS can produce very competitive performance, as the linux folks have seen from time to time, especially in the enterprise space.
Also true. There is much to be learned from the kids in Redmond. I remember reading Linus' postings a while after the second round of tests that showed Win2K beating Linux. Basically he said, OK, they beat us fair and square. We've got lots of work to do, let's get to it.
you have to remember that apple cannot afford to put all its eggs in intel products right now, once they decide to go intel intel has to commit in a certain way to support apple.
this means that there is still a significant level of investment (r&d, marketing, etc) in the powerpc product pipeline. i feel if they released intel-macs right now (even if it were possible), that would mean a lot of investment in powerpc goes straight into the trash.
You haven't convinced me yet. Effectively that investment is already in the trash, or more properly those are sunk costs. For the most part Apple isn't going to be able to sell both a PPC and a Mactel to the same customer in the span of 1.5 years. There is the very real risk that many of their customers will wait until they can buy an intel box, so releasing those boxes sooner rather than later mitigates that risk and gives Apple the best chance to minimize the losses of the transition. The only explanation for the delay that makes sense to me so far is that the translated performance really isn't all that great, and Apple is afraid that that will cause an even larger exodus than previous transitions have. Thus, they have to wait until native binaries are available and Rosetta really only exists to cover shareware and older, non-critical apps.
Quote:
with the current scenario.
1. they can recoup their investments on powerpc line by clearing their pipeline of powerpc Macs over next 1-2 years
2. they have powerPC fallback while intel Macs are developed
3. apple and intel have 6-12 months from today to deliver on intel-macs, this has to be pretty much flawless or apple is seriously screwed compared to xp/longhorn/linux on intel
4. they have to give time for the developers to come to grips with this
5. rosetta is very important but i agree, they have decided this is not enough. by apple's standards, fat binaries MUST be employed as far as possible*
*i think the core reason why rosetta won't cut it for the most part is because comparative benchmarks of many applications running on similarly spec'ed intel hardware may show windows/linux versions of those applications with a clear advantage over mac os X-rosetta'ed apps.
1. See above.
2. Apple must have been thinking about this for at least a few months. They should be able to produce volume units by the end of the year unless there are significant issues with locking down X-on-apple only. I doubt that is the case.
3. That's a risk for the rest of time now, and more on the software side than the hardware.
4-5. If Rosetta is really that fantastic, then they don't have to wait. As I said before, my hypothetical $1 is that Rosetta has been implicitly overstated.
Comments
Originally posted by sillyfool
Over the past 15 years the UNIX world has gone through this same conversion. I can't tell you the number of times I've heard all these same things said by the UNIX geeks as their favorite botique architectures have been replaced by x86 at the desktop, workstation, and small-to-mid sized server levels.
Well, except that all of those companies that transitioned are gone (DEC) or are mere shells of their former selves (SGI, Sun).
I agree with your overall sentiment, but I don't think that point is strong...
After milling this over two fitful sleepless nights (why can worrying about a f**ing computer do this to me), I am content and at peace with Apple's decision. It is the best. Are there risks? Of course.
The fact of the matter is that Apple could no longer be in the CPU/Bus Design business. It is FAR better to use the resources the rest of the world rely on. Altivec? Who cares. I used to, but after doing Altivec optimizations for my model (what a pain), I have realized that I would rather have a fast straight-up FPU. The G5 and AMD's have it now, the Intel chips coming out are rumored to surpass them both with ease. This leads to another point. Rumors regarding Intel chips tend to come true. The number of unfulfilled rumors, hope, dreaming, speculation from PPC the past 7 years have been frustrating with only a SINGLE breath of fresh air (G5).
The compilers, optimization libraries, etc. available on the x86 platform are incredible. Hell, we were all excited because Tiger was compiled with gcc 3.4 with more PPC optimization built-in. Building this support was hard work for Apple, now it is free.
Maybe, just maybe, Intel needs Apple. Intel has been at the mercy of Microsoft for too long now. Intel has always been Microsoft's biggest ally/enemy. The prepondance of cheap hardware with the ubiquity of Windows has built a great empire for them both; however, Microsoft is the more powerful as they are software only (excluding the xbox). The threats and manipulations that MS have placed on Intel over the years were exposed during the antitrust trial. Intel has and remains beholden to Microsoft's wishes. Intel thought that Netscape, Java, etc. might be their way out and they were put back into their place.
Intel has been pushing Linux, and will now presumably be an ally of MacOS X because they need software support from someone other than MS. I wouldn't be surprised if Intel gave Apple a big discount on the chips. Hell, they may have given Apple Dell like prices.
Well, except that all of those companies that transitioned are gone (DEC) or are mere shells of their former selves (SGI, Sun).
The market cap. of HP/Compaq is more than twice (almost 3 times) that of Apple.
Sun Microsystems is at least 1.5x of Apple.
I agree with your overall sentiment, but I don't think that point is strong
I think that you missed my point. My point was not that the vendors of boutique MPU architectures were doing well in the market. My point was that the market had compelled the UNIX world to abandon those architectures and transition to x86.
Originally posted by sillyfool
The market cap. of HP/Compaq is more than twice (almost 3 times) that of Apple.
Sun Microsystems is at least 1.5x of Apple.
Well, to be annoying...
The market cap of Sun is 12B versus 30B for Apple. SGI is 214M. HP's market cap is 64B; however, just because Compaq bought DEC, I don't consider them the same company...
I did miss your point. You are exactly right, the market forced all of these companies (well, Sun still does SPARC development) to stop financing their own CPU's.
Originally posted by atomicham
well, Sun still does SPARC development
But probably not for much longer. I can imagine them going the Apple route (they already started down that path actually).
What makes all of this that much more interesting is the position it puts/keeps Intel in..."Monopolist to the Stars".
Of course both Apple and Sun are smart enough to keep the CPU hidden enough that switching around becomes much easier (Apple is probably doing this better than Sun). Either way, it seems that MS is in a much moer precarious position that way.
Intel would love to see Apple become more successful from a marketshare perspective. It means that there is a more diverse OS base running on their chip. Not just Windows. They must love Linux and Solaris too.
I think there could be some long term problems for MSFT.
Everyone is keeping options all over the place. Microsoft did this back before their OS dominance. One wonders if they are now (e.g., "secret double life for Office...running on Linux") or more concerned about protecting current business.
-BUBBA
The market cap of Sun is 12B versus 30B for Apple.
Yup, sorry about that. I read today's volume number not today's market cap.
And to be fair, I should have said that the MPU architectures for these vendors had been forced out of the merchant mobile/desktop/workstation/small-to-mid range server space.
Alpha is end-of-life; no designs for any applications.
PA-RISC is end-of-life; no designs for any applications.
MIPS still exists in the embedded space but there are no new MPU designs for non-embedded applications.
SPARC has been abandoned by Sun. Fujitsu will continue to develop newer generation SPARC derivative MPUs. It simply became too expensive for Sun to push Sparc forward and to work on the next generation architecture.
Originally posted by sillyfool
The path to where we are is strewn with the empty husks of the 'better' architectures that have fallen under the x86 boot. Alpha, MIPS, PA-RISC, SPARC, good bye and thanks for all the fish.
Saddest (tech) sentence I've read this year.
Applies to BeOS, OS/2, Oberon, Plan 9, etc under the heel of Windows, too.
There is no logic or justice in the world. Let's hope OS X doesn't end up as a husk as well.
Originally posted by kwatson
Saddest (tech) sentence I've read this year.
Applies to BeOS, OS/2, Oberon, Plan 9, etc under the heel of Windows, too.
There is no logic or justice in the world. Let's hope OS X doesn't end up as a husk as well.
don't be so sad, dude. we as geeks have been programmed in the 90s to think that RISC is da bomb and CISC is dead... but such things are not as important in the next 10 years. in the next 10 years, it will be squeezing as much as possible out of 90nm, 65, 45?nm to deliver the cost, power, lowheat, etc. consumers demand ~ this will also translate upwards into the server space to some degree...
in 10 years time RISC vs x86 or whatever will be irrelevant with quantum, molecular, biological, nanotech, optical or whatever to bring us on.
we KNOW mac os X has a great future ahead of it, at least 5 years, and people are getting tired of windows, they just don't have any options. 1-10 years out, Intel seems to offer apple what it needs, and yes, i agree with previous posters that Intel can get out from under microsoft's thumb. linux-on-intel, macosX-on-intel, should be alright... mac os X-on-powerPC, well, i am mourning it now, but it's got a few more years left in it, a very safe, stable bet for at least 1-2 years until the dust settles.
... and people are getting tired of windows, they just don't have any options.
Linux has come a very long way. Almost every company that I know of have Windows and Linux boxen but zero Macs.
is far superior to ANY operating system in the world.
The people still defending Microsoft are simply over confident about there job security.
Now we have the maker of 80% of the words processors telling the entire industry that Mac OS X IS clearly the superior OS to ANYTHING.
Originally posted by FallenFromTheTree
Every review I've read about this says that LInux is O.K. but OSX
is far superior to ANY operating system in the world.
The people still defending Microsoft are simply over confident about there job security.
Now we have the maker of 80% of the words processors telling the entire industry that Mac OS X IS clearly the superior OS to ANYTHING.
We'll see what kind of song they're singing as Longhorn and Leopard approach.
are stuck in a big nasty fly infested pile of manure
Every review I've read about this says that LInux is O.K. but OSX
is far superior to ANY operating system in the world.
Spend some more time on Slashdot. And Anandtech can be usefull some times: http://anandtech.com/mac/showdoc.aspx?i=2436
To quote Johan De Gelas ( of Aceshardware.com fame ), from the final section of that review:
The server performance of the Apple platform is, however, catastrophic. When we asked Apple for a reaction, they told us that some database vendors, Sybase and Oracle, have found a way around the threading problems. We'll try Sybase later, but frankly, we are very sceptical. The whole "multi-threaded Mach microkernel trapped inside a monolithic FreeBSD cocoon with several threading wrappers and coarse-grained threading access to the kernel", with a "backwards compatibility" millstone around its neck sounds like a bad fusion recipe for performance.
Originally posted by Macfr3ak
Ok, how about the CELL? Is IBM so much incompetent? In your way of thinking, why Apple suddenly left the PowerPC "Boat"?
Cell's first implementation is designed for a game console and does not deliver stellar performance on non-vectorized code. Its performance is not terrible, but obviously Apple wants top-notch performance on typical code. This doesn't mean IBM is incompetent, it means they have other goals (i.e. selling many millions of chips in consumer units). That is a distraction from what Apple needs, and that is something I've been afraid of since their involvement in Cell & XBox360 first became known.
I had been hoping that a middle ground could be reached and a Cell variant suited to Apple would keep Apple on the PPC course, but I have my own set of predjudices.
Do you think this company is unaware of it? Don't you think INTEL has more tools for instance to lock some toys with DRM'ized Chipsets?
You don't know what DRM support IBM might be providing in Cell.
PS: Sorry Programmer, Personnely, I'm very upset with Apple's hardware's choice.
Maybe I'm wrong? who knows.....
Unfortunately IBM's main business goals pretty much forced them into it.
Originally posted by melgross
I was going to mention the SIMD at some point.
What we have to understand here is that Intel is not the dog uninventive company that some think it is. As Ottelini (spelling?) shower, for those who might not have known, Intel inventor the DRAM, and the microprocessor itself.
Assuming you're talking about 1T cells, then I think IBM would have something to say about that. If you want to talk about DRAM in general, then you definitely cannot credit that to intel. They can claim to have the first commercial product and, of course, they get credit for the 4004.
Quote:
They were one of the originators of Flash memory and are the biggest supplier of that as well.
Again, I'm assuming that you're not limiting the discussion to NOR flash in which case that honor goes to Samsung
Quote:
Over the years Intel made several contributions to cpu design only to find MS in opposition to those innovations because they wanted to do it in software. Their software.
Intel came out with several SIMD extensions that were never used, and were therefore dropped. The ones in place now are the ones that MS supports, or work without software support.
If Intel was willing to take the risks over the years of coming out with innovations that they didn't know MS would support, they might be interested in doing that again. But this time with the support of Apple.
If this happens, and it's a greater possibility than some of what I've been reading here and elsewhere, then MS would feel compelled to support them as well, or lose performance to an x86 Mac OS.
Intel might be very happy at this turn of events because it will allow the to regain control of all of their chip development.
Who knows, this could be one of the reasons they've been wooing Apple for so long.
Intel has clearly been able to support their ISA for far longer than many have predicted, and they've done it with some very clever hacks. While I'm sure that Intel will continue to add instructions and functionality to their chips, MS certainly won't be intimidated by having new functions available. The only real threat to MS would be if Intel were to abandon x86 in favor of a more modern and elegant ISA. This simply isn't going to happen for a very long time, so this isn't really a credible threat and does not unencumber Intel from that burden.
Frankly what suprises me most about Jobs' statements is the comment about performance/W. Intel has to burn a decent amount of power to simply handle the instruction decoding which was always a clear advantage to PPC. The only interpretation of this that makes sense to me is in the portable space where PPC obviously hasn't kept pace. The other possibility I see is that Apple is looking to make more home media devices and believes that Intel has some rabbits in that area, but given PPC's successes in the embedded market, I'm skeptical.
I agree with previous posters that this announcement and the timeline laid out by Apple will significantly hurt Mac sales. Actually I predict that there will be a spike in sales followed by a drawn out lull as certain groups load up on the hardware they need for the next 3-4 years, including spares, followed by the unwashed masses not wanting to invest in a lame duck platform.
Finally, has anyone else noticed the inconsistency between Rosetta and a very late introduction of the first Mactels? If Rosetta is indeed that great, then why wouldn't Apple be releasing an Intel box before the end of the year just to mitigate any sales losses and to seriously refresh their powerbooks? Or is binary translation really not quite the panacea Jobs makes it out to be (shocking to think that Stevie might be exaggerating I know!), and in order for this switch to work Apple desperately needs the developers to make native binaries. My hypothetical money's on the latter.
Originally posted by Programmer
......
I had been hoping that a middle ground could be reached and a Cell variant suited to Apple would keep Apple on the PPC course.....
a sentiment and hope echoed by many, but sadly, appears to be dashed.
additionally,
my guess is that it looks like i-series/power-series Mac
h.264 hardware-based decode/encode(?) will be handled by intel/ati, very unlikely to be Cell like many were hoping for
Originally posted by urp
.......
Finally, has anyone else noticed the inconsistency between Rosetta and a very late introduction of the first Mactels? If Rosetta is indeed that great, then why wouldn't Apple be releasing an Intel box before the end of the year just to mitigate any sales losses and to seriously refresh their powerbooks? Or is binary translation really not quite the panacea Jobs makes it out to be (shocking to think that Stevie might be exaggerating I know!), and in order for this switch to work Apple desperately needs the developers to make native binaries. My hypothetical money's on the latter.
you have to remember that apple cannot afford to put all its eggs in intel products right now, once they decide to go intel intel has to commit in a certain way to support apple.
this means that there is still a significant level of investment (r&d, marketing, etc) in the powerpc product pipeline. i feel if they released intel-macs right now (even if it were possible), that would mean a lot of investment in powerpc goes straight into the trash.
with the current scenario.
1. they can recoup their investments on powerpc line by clearing their pipeline of powerpc Macs over next 1-2 years
2. they have powerPC fallback while intel Macs are developed
3. apple and intel have 6-12 months from today to deliver on intel-macs, this has to be pretty much flawless or apple is seriously screwed compared to xp/longhorn/linux on intel
4. they have to give time for the developers to come to grips with this
5. rosetta is very important but i agree, they have decided this is not enough. by apple's standards, fat binaries MUST be employed as far as possible*
*i think the core reason why rosetta won't cut it for the most part is because comparative benchmarks of many applications running on similarly spec'ed intel hardware may show windows/linux versions of those applications with a clear advantage over mac os X-rosetta'ed apps.
Originally posted by sillyfool
Spend some more time on Slashdot. And Anandtech can be usefull some times: http://anandtech.com/mac/showdoc.aspx?i=2436
To quote Johan De Gelas ( of Aceshardware.com fame ), from the final section of that review:
The server performance of the Apple platform is, however, catastrophic. When we asked Apple for a reaction, they told us that some database vendors, Sybase and Oracle, have found a way around the threading problems. We'll try Sybase later, but frankly, we are very sceptical. The whole "multi-threaded Mach microkernel trapped inside a monolithic FreeBSD cocoon with several threading wrappers and coarse-grained threading access to the kernel", with a "backwards compatibility" millstone around its neck sounds like a bad fusion recipe for performance.
Agreed. Even IBM realized that mach was a waste of time and energy. Apple lost me, but not my wife, with the switch to X. I was never a fan of the nExt architecture and by extension X, and the few interface gains in X were offset by the losses. X could and still can't make its mind up. Is it a unix box? Is it mach? All of that has limited what could be done, and now that the hardware will be identical does open Apple up to some potentially unflattering comparisons. For all the talk in this thread about giving credit to Intel for producing competitive hardware, everyone should be willing to admit that MS can produce very competitive performance, as the linux folks have seen from time to time, especially in the enterprise space.
Originally posted by urp
Agreed. Even IBM realized that mach was a waste of time and energy...
Yup, me and Mach both left CMU at about the same time, so I had a bit of a soft spot for it. But facts-is-facts.
For all the talk in this thread about giving credit to Intel for producing competitive hardware, everyone should be willing to admit that MS can produce very competitive performance, as the linux folks have seen from time to time, especially in the enterprise space.
Also true. There is much to be learned from the kids in Redmond. I remember reading Linus' postings a while after the second round of tests that showed Win2K beating Linux. Basically he said, OK, they beat us fair and square. We've got lots of work to do, let's get to it.
Originally posted by sunilraman
you have to remember that apple cannot afford to put all its eggs in intel products right now, once they decide to go intel intel has to commit in a certain way to support apple.
this means that there is still a significant level of investment (r&d, marketing, etc) in the powerpc product pipeline. i feel if they released intel-macs right now (even if it were possible), that would mean a lot of investment in powerpc goes straight into the trash.
You haven't convinced me yet. Effectively that investment is already in the trash, or more properly those are sunk costs. For the most part Apple isn't going to be able to sell both a PPC and a Mactel to the same customer in the span of 1.5 years. There is the very real risk that many of their customers will wait until they can buy an intel box, so releasing those boxes sooner rather than later mitigates that risk and gives Apple the best chance to minimize the losses of the transition. The only explanation for the delay that makes sense to me so far is that the translated performance really isn't all that great, and Apple is afraid that that will cause an even larger exodus than previous transitions have. Thus, they have to wait until native binaries are available and Rosetta really only exists to cover shareware and older, non-critical apps.
with the current scenario.
1. they can recoup their investments on powerpc line by clearing their pipeline of powerpc Macs over next 1-2 years
2. they have powerPC fallback while intel Macs are developed
3. apple and intel have 6-12 months from today to deliver on intel-macs, this has to be pretty much flawless or apple is seriously screwed compared to xp/longhorn/linux on intel
4. they have to give time for the developers to come to grips with this
5. rosetta is very important but i agree, they have decided this is not enough. by apple's standards, fat binaries MUST be employed as far as possible*
*i think the core reason why rosetta won't cut it for the most part is because comparative benchmarks of many applications running on similarly spec'ed intel hardware may show windows/linux versions of those applications with a clear advantage over mac os X-rosetta'ed apps.
1. See above.
2. Apple must have been thinking about this for at least a few months. They should be able to produce volume units by the end of the year unless there are significant issues with locking down X-on-apple only. I doubt that is the case.
3. That's a risk for the rest of time now, and more on the software side than the hardware.
4-5. If Rosetta is really that fantastic, then they don't have to wait. As I said before, my hypothetical $1 is that Rosetta has been implicitly overstated.