By the way, none of these folks were involved during the Apple merger with NeXT, so speaking to a group of Apple employees and not NeXT Employees is a waste of time.
More to the point, Apple had multiple projects from Blue Box, Red Box [Windows], Yellow Box [Cocoa on NT they acquired from NeXT], TaligentOS that was originally positioned against NeXTStep/Openstep and BeOS, to many other projects, including MkLinux.
In fact, we used MKLinux to bootstrap on Openstep to PowerPC 604/620 originally while modifying the Mach microkernel, the toolkit and more to get a native bootloading process for PowerPC.
We already had m68k so it wasn't a huge process--much of the work spent was introducing Apple Engineers to the way of NeXT Design and Thinking. NeXT Thought Different than Apple in every facet.
Interview Bertrand Serlet and Avi Tevanian, and then I'll be impressed. Jean-Louis Gassée's arrogance was so hilarious when he demanded $100 million for himself personally plus twice what Apple paid for NeXT before he'd sell BeOS to Apple.
I remember there being a debate at the time about the merits of PowerPC's RISC processor architecture vs. Intel's CISC architecture.
Some in the pro-PowerPC camp liked that Apple used an exotic CPU, and that it added to the appeal of Macs being different (meaning, better). And the pro-CISC camp hoping a move to Intel would make CPUs cheaper for Apple.
In hindsight, it's pretty clear the move to Intel CPUs made Macs much more appealing for PC switchers, being able to dual-boot and also having access to more easily ported Windows games. From a marketing standpoint, it also let Macs go apples-to-apples with Windows machines, and helped people see that eventhough they have the same CPU inside, Macs performed better than Windows machines.
I'm glad they made the change. Although I'm still computing on a G5 machine, I look forward to my next upgrade to an Intel iMac, which will let me have the advantages mentioned above.
Hello RISC, your ARM sure looks fantastic these days.
Even your hybrid CISC/RISC in Intel is quite sexy.
How much of NeXT-created software actually made it into the Mac? I read somewhere that the Mach kernel was the main thing ported over, but that wasn't made by NeXT.
Pretty much the entire OS was brought over. Cocoa, Core Foundation, etc, were all developed at NeXT.
So, they got together a panel with Tesler, the guy who screwed up the future System 8 so badly that Apple had to buy another company (I had friends in Apple back then, ATG was a huge problem); the Frenchman and former incompetent Apple manager who wanted to sell his company to Apple but had an OS just as useless as System 8; Hertzfeld, who was really talented, but long gone before any of this happened; Atkinson, same thing.
And they make the ridiculous claim that Apple was looking at switching to Intel's slow, messy architecture? Seriously?
Back then, the PPC could run circles around anything Intel had. Even Intel was looking at abandoning it, what was to become Itanium was to be the future for Intel. And it wasn't until IBM failed to deliver low-power G5 chips suitable for laptop use that there was ever any real need or reason at all to switch. The PPC remained a better desktop chip right up until Intel came out with the Core series, the P4 was garbage.
No, Apple didn't buy NeXT because they wanted to switch to Intel. That could have happened anytime after the NeXT purchase, since that code was already written and it took extra effort to port to PPC, NeXTSTEP was M68k and Intel, never PPC. It turned out that keeping the Intel code going as a contingency was a good idea, so they did that. But only a fool would have switched any earlier than Apple did.
I'm betting this bogus story is primarily Tesler, since his group was responsible for a good chunk of the failure to write a next-generation operating system for the Mac.
The PPC remained a better desktop chip right up until Intel came out with the Core series, the P4 was garbage.
Actually, the P4-based systems Apple provided to developers in 2005 knocked the socks off PowerMacs for performance in general purpose computing. Intel probably even surprised itself 10 years earlier with the Pentium Pro, which was quite simply a great processor that possibly put a lot of competitive pressure on the PowerPC consortium.
As for instruction sets? Yes, X86 sucks big time and the P4 power envelope was awful. It's too bad Apple needed to transition to Intel so soon that X86 had to be supported and a clean jump to X64 couldn't be made. But Apple did a superb job nonetheless!
How much of NeXT-created software actually made it into the Mac? I read somewhere that the Mach kernel was the main thing ported over, but that wasn't made by NeXT.
Quite a lot. The Kernel, most of the open sourced Unix stuff, the dock, finder, the display manager (originally based on Display PostScript from Adobe, later ported to Display PDF on OS X) and the Cocoa libraries came from NeXT. Apple ported it all, added the now-obsolete Carbon libraries to simplify porting of Mac OS 9 applications, and of course, Aqua was new. There was definitely porting work; Apple didn't start from scratch. Think of OS X being more NeXTStep version 5 and less Mac OS version 10.
Going with next was the best decision apple of the time made. It turned out to be quite a flexible platform for both desktop, phone and even tablet os. I wonder if they could have scaled beos the same way.
Sounds like it was a best decision, still would be interesting to speculate how Apple would fare if they made a deal with Be instead. Would they survive ? And what about NeXT and Jobs ? Interesting I never found any such speculative article. There is quite informative info on Be vs. NeXT decision :
"If you refuse to update your software for six years, you obviously don't care about your customers and you deserve to be artificially obsoleted."
Something like that, perhaps?
Your statements are often bold and wrong.
Case in point, Blizzard, which had spent money for years maintaining its PowerPC for their customers. They are the _only_ game company to behave which such care for their customers, and everyone knows how much Apple cares for its older software/customers/machines. Addendum, do you have the slightest idea of the cost of reprogramming a game (or any piece of software) for a new architecture? Unless you can get people to pay again...
Hint: Apple had to buy next because it was so hard to translate Mac OS to a new architecture!
Conclusion: it's a matter of business, not of "deserving".
I wished at the time that Apple would simply have bought the rights to the Amiga OS, which was multitasking and stable from day 1, but since Motorola fell behind Intel things turned out alright
Addendum, do you have the slightest idea of the cost of reprogramming a game (or any piece of software) for a new architecture?
An operating system, yes that would be an absolute pain! As would software that had to tap directly into the hardware.
Software programmed in a cross platform language that tap into common APIs make its as simple as copying the source code from the PPC machine over to the Intel machine and hitting the 'compile' button.
Quote:
Originally Posted by lightknight
Your statements are often bold and wrong.
Case in point, Blizzard, which had spent money for years maintaining its PowerPC for their customers
No, you misunderstand completely. Blizzard have PPC for legacy machines and Intel for newer machines. Tallest Skil Was saying if you only have PPC software and no Intel alternative then you deserve to have a nice kick in the teeth for not updating the software in six years.
It's amazing how much you can forget. I have been using Macs since their intro, and have every version of Mac OS from 6 onwards (actually, I have every OS Apple has ever produced going back to the garage days), and followed Apple's moves every day as I still do today.
I think that Apple put a lot of pressure on Intel to improve its processors, on a promise that all future Macs would use them. Just look at the Apple-driven motivation for "miniaturising" of the processor for Macbook Air -- one reason that I find compelling enough for Apple to stay with Intel instead of migrating the Air to the A5/A6 processors -- but, hey! I'm not Steve Jobs so what do I know?
...because it didn't happen that way. These guys are talking about events that happened in the 1996 timeframe. In 1996, the PowerPC had a bright future. Apple had transitioned to PowerPC just a couple of years prior. The G3 was still two years off. Intel was still looking for a graceful way to dump IA-32 in favor of a more modern processor architecture.
Intel processors of that era were HOT! Apple did not switch to Intel until a decade later. One major reason was the Intel processors of the mid-2000s were a lot cooler than the PowerPCs of the era. The other major reason was that IBM refused to design cooler versions of the PowerPC. Among other things, this meant that a PowerBook G4 was hot and that the PowerBook G5 was impossible to build.
I have no doubt that Apple had many serious discussions and studies of many contingencies. The fact that Apple switched to Intel ten years after Steve Jobs returned to the company is strong evidence that switching processors was not a high priority for most of that time.
Quote:
Originally Posted by nagromme
I agree, the need to be able to run on Intel can?t have been the whole story. But it?s interesting to know how early on it was even PART of the story!
(In fact, if Intel had never gotten their act together post-Pentium-4, and if IBM and Motorola hadn?t abandoned personal computer processor development, it may well never have made sense to switch to Intel at all. The option to be ABLE to was certainly important, though, and was an option Apple had long been working on.)
If this contributed to the NeXT choice and the return of Steve, then it contributed to OS X, iPod, iOS, iCloud, every Apple product line sold today, AND therefore, the shape of the computer AND phone AND music industries today, as well as the entire touch-computing paradigm that has begun slowly replacing the mouse for everyday computing! It made Android possible (as more than the Blackberry clone it was pre-iPhone). Even Windows Phone 7: it?s not an iPhone-mimic like Android, but without iPhone, and without iPhone causing the rise of Android, it?s likely Microsoft would never have developed Windows Phone 7 as we know it.
Quote:
Originally Posted by karmadave
I really think there is more to it than 'we needed an OS that could run on Intel'. Apple spent years trying to develop a modern OS, to replace the original MacOS and failed! At that point, their ONLY option was to acquire an OS and put their UI on it. In addition to NeXT and Be, Apple was looking at other Unix variants such as Sun Solaris which ran on SPARC, Intel, and PowerPC. They ultimately chose NeXT, but other than the fact that NeXTStep (renamed 'OpenStep') was being de-emphasized in favor of NetObjects, Tesler's comments do NOT tell the whole story...
I very much doubt Apple bought NeXT with the express intent to switch to the Intel architecture, but it makes sense that they shopped specifically for a portable OS. It doesn't get much more portable than NeXT: it ran on four architectures when Apple bought it. it should come as no surprise that Apple closely guarded that portability from the very beginning, and I'm quite confident they're maintaining it on different architectures to this very day, keeping all the mechanisms for a swift and relatively smooth transition in place.
Copland was a project at Apple Computer to create an updated version of the Macintosh operating system. It was to have introduced protected memory, preemptive multitasking and a number of new underlying operating system features, yet still be compatible with existing Mac software. ...
which coincides with what I recall actually going on at the time.
The early '90s was the period when preemptive multitasking and a number of other features we now take for granted were introduced to personal computer operating systems. Copland's stillbirth was a major problem for Apple, and, at least in retrospect, symptomatic of much bigger problems at the company.
I think this story about interest in an Intel switch being the reason for acquiring NeXT is a rewriting of history. Apple desperately needed a modern operating system and the fact that NeXTSTEP was already running on Intel was just a bonus for them, one that wasn't squandered.
Rosetta was the second emulator that Apple had to implement. The first being the 68K emulator that was part of Classic.
I believe that Rosetta involved a considerable number of hooks in OS X to make it work correctly. EOL'ing Rosetta cleaned up the build tree I'm sure. At this point they knew the remaining number of non-UB apps was minimal, and there were sufficient replacements for the few that never transitioned.
Pity they never re-compiled AppleWorks 6 as a UB. I wonder if this was a business decision (in light of iWorks) or if AW6 had sufficient non-portable code down deep within. I would care a lot less about this one, had Apple implemented a proper database module as part of iWorks (ignoring Bento for this discussion). Anyone with legacy data in AW6 is stuck at 10.6.8.
BTW, this article finally adds color to where/when Apple soured on PPC. Up until now, most people thought it happened after the NeXT acquisition. Now that we see it goes back much farther, then it appears to be a matter of timing. The promised G5 at 3 Ghz, which IBM missed the deadline on, was probably the tipping point that pushed the decision over the edge.
I suspect that Apple has a new Rosetta that allows x86 compiled app to run on ARM
Rosetta only worked on i86 because PPC was such a terrible professor. I said so in the forums years ago and people ignored me. The fact I'd PPC processors had terrible integer performance while intel was many times better. This additional integer capacity allowed for impressive emulation of PPC code.
ARM is already at a deficit with respect to intel hardware so any sort of Intel emulation will be pretty pathetic. The only way ARM will be viable near term is if the machine is running native code or hot compiling code with the facilities of LLVM. If not native performance would more than suck.
I suspect that Apple has a new Rosetta that allows x86 compiled app to run on ARM
I doubt that. But an ARM/iOS based desktop (similar to the iMac) is more likely. That way, the switchers who are buying iOS devices at a fast clip can run the same app on a home based desktop. The gestures would have to be done via the Magic Trackpad tho.
Intel processors of that era were HOT! Apple did not switch to Intel until a decade later. One major reason was the Intel processors of the mid-2000s were a lot cooler than the PowerPCs of the era. The other major reason was that IBM refused to design cooler versions of the PowerPC. Among other things, this meant that a PowerBook G4 was hot and that the PowerBook G5 was impossible to build.
Don't forget that the G5 (the last few used in the Power Mac) had to be water cooled. That is the only Apple product ever released that needed that level of cooling.
Comments
More to the point, Apple had multiple projects from Blue Box, Red Box [Windows], Yellow Box [Cocoa on NT they acquired from NeXT], TaligentOS that was originally positioned against NeXTStep/Openstep and BeOS, to many other projects, including MkLinux.
In fact, we used MKLinux to bootstrap on Openstep to PowerPC 604/620 originally while modifying the Mach microkernel, the toolkit and more to get a native bootloading process for PowerPC.
We already had m68k so it wasn't a huge process--much of the work spent was introducing Apple Engineers to the way of NeXT Design and Thinking. NeXT Thought Different than Apple in every facet.
Interview Bertrand Serlet and Avi Tevanian, and then I'll be impressed. Jean-Louis Gassée's arrogance was so hilarious when he demanded $100 million for himself personally plus twice what Apple paid for NeXT before he'd sell BeOS to Apple.
What a douche.
I remember there being a debate at the time about the merits of PowerPC's RISC processor architecture vs. Intel's CISC architecture.
Some in the pro-PowerPC camp liked that Apple used an exotic CPU, and that it added to the appeal of Macs being different (meaning, better). And the pro-CISC camp hoping a move to Intel would make CPUs cheaper for Apple.
In hindsight, it's pretty clear the move to Intel CPUs made Macs much more appealing for PC switchers, being able to dual-boot and also having access to more easily ported Windows games. From a marketing standpoint, it also let Macs go apples-to-apples with Windows machines, and helped people see that eventhough they have the same CPU inside, Macs performed better than Windows machines.
I'm glad they made the change. Although I'm still computing on a G5 machine, I look forward to my next upgrade to an Intel iMac, which will let me have the advantages mentioned above.
Hello RISC, your ARM sure looks fantastic these days.
Even your hybrid CISC/RISC in Intel is quite sexy.
How much of NeXT-created software actually made it into the Mac? I read somewhere that the Mach kernel was the main thing ported over, but that wasn't made by NeXT.
Pretty much the entire OS was brought over. Cocoa, Core Foundation, etc, were all developed at NeXT.
So, they got together a panel with Tesler, the guy who screwed up the future System 8 so badly that Apple had to buy another company (I had friends in Apple back then, ATG was a huge problem); the Frenchman and former incompetent Apple manager who wanted to sell his company to Apple but had an OS just as useless as System 8; Hertzfeld, who was really talented, but long gone before any of this happened; Atkinson, same thing.
And they make the ridiculous claim that Apple was looking at switching to Intel's slow, messy architecture? Seriously?
Back then, the PPC could run circles around anything Intel had. Even Intel was looking at abandoning it, what was to become Itanium was to be the future for Intel. And it wasn't until IBM failed to deliver low-power G5 chips suitable for laptop use that there was ever any real need or reason at all to switch. The PPC remained a better desktop chip right up until Intel came out with the Core series, the P4 was garbage.
No, Apple didn't buy NeXT because they wanted to switch to Intel. That could have happened anytime after the NeXT purchase, since that code was already written and it took extra effort to port to PPC, NeXTSTEP was M68k and Intel, never PPC. It turned out that keeping the Intel code going as a contingency was a good idea, so they did that. But only a fool would have switched any earlier than Apple did.
I'm betting this bogus story is primarily Tesler, since his group was responsible for a good chunk of the failure to write a next-generation operating system for the Mac.
Wow.
The PPC remained a better desktop chip right up until Intel came out with the Core series, the P4 was garbage.
Actually, the P4-based systems Apple provided to developers in 2005 knocked the socks off PowerMacs for performance in general purpose computing. Intel probably even surprised itself 10 years earlier with the Pentium Pro, which was quite simply a great processor that possibly put a lot of competitive pressure on the PowerPC consortium.
As for instruction sets? Yes, X86 sucks big time and the P4 power envelope was awful. It's too bad Apple needed to transition to Intel so soon that X86 had to be supported and a clean jump to X64 couldn't be made. But Apple did a superb job nonetheless!
How much of NeXT-created software actually made it into the Mac? I read somewhere that the Mach kernel was the main thing ported over, but that wasn't made by NeXT.
Quite a lot. The Kernel, most of the open sourced Unix stuff, the dock, finder, the display manager (originally based on Display PostScript from Adobe, later ported to Display PDF on OS X) and the Cocoa libraries came from NeXT. Apple ported it all, added the now-obsolete Carbon libraries to simplify porting of Mac OS 9 applications, and of course, Aqua was new. There was definitely porting work; Apple didn't start from scratch. Think of OS X being more NeXTStep version 5 and less Mac OS version 10.
Going with next was the best decision apple of the time made. It turned out to be quite a flexible platform for both desktop, phone and even tablet os. I wonder if they could have scaled beos the same way.
Sounds like it was a best decision, still would be interesting to speculate how Apple would fare if they made a deal with Be instead. Would they survive ? And what about NeXT and Jobs ? Interesting I never found any such speculative article. There is quite informative info on Be vs. NeXT decision :
http://macspeedzone.com/archive/art/con/be.shtml
"If you refuse to update your software for six years, you obviously don't care about your customers and you deserve to be artificially obsoleted."
Something like that, perhaps?
Your statements are often bold and wrong.
Case in point, Blizzard, which had spent money for years maintaining its PowerPC for their customers. They are the _only_ game company to behave which such care for their customers, and everyone knows how much Apple cares for its older software/customers/machines. Addendum, do you have the slightest idea of the cost of reprogramming a game (or any piece of software) for a new architecture? Unless you can get people to pay again...
Hint: Apple had to buy next because it was so hard to translate Mac OS to a new architecture!
Conclusion: it's a matter of business, not of "deserving".
Addendum, do you have the slightest idea of the cost of reprogramming a game (or any piece of software) for a new architecture?
An operating system, yes that would be an absolute pain! As would software that had to tap directly into the hardware.
Software programmed in a cross platform language that tap into common APIs make its as simple as copying the source code from the PPC machine over to the Intel machine and hitting the 'compile' button.
Your statements are often bold and wrong.
Case in point, Blizzard, which had spent money for years maintaining its PowerPC for their customers
No, you misunderstand completely. Blizzard have PPC for legacy machines and Intel for newer machines. Tallest Skil Was saying if you only have PPC software and no Intel alternative then you deserve to have a nice kick in the teeth for not updating the software in six years.
I think that Apple put a lot of pressure on Intel to improve its processors, on a promise that all future Macs would use them. Just look at the Apple-driven motivation for "miniaturising" of the processor for Macbook Air -- one reason that I find compelling enough for Apple to stay with Intel instead of migrating the Air to the A5/A6 processors -- but, hey! I'm not Steve Jobs so what do I know?
...because it didn't happen that way. These guys are talking about events that happened in the 1996 timeframe. In 1996, the PowerPC had a bright future. Apple had transitioned to PowerPC just a couple of years prior. The G3 was still two years off. Intel was still looking for a graceful way to dump IA-32 in favor of a more modern processor architecture.
Intel processors of that era were HOT! Apple did not switch to Intel until a decade later. One major reason was the Intel processors of the mid-2000s were a lot cooler than the PowerPCs of the era. The other major reason was that IBM refused to design cooler versions of the PowerPC. Among other things, this meant that a PowerBook G4 was hot and that the PowerBook G5 was impossible to build.
I have no doubt that Apple had many serious discussions and studies of many contingencies. The fact that Apple switched to Intel ten years after Steve Jobs returned to the company is strong evidence that switching processors was not a high priority for most of that time.
I agree, the need to be able to run on Intel can?t have been the whole story. But it?s interesting to know how early on it was even PART of the story!
(In fact, if Intel had never gotten their act together post-Pentium-4, and if IBM and Motorola hadn?t abandoned personal computer processor development, it may well never have made sense to switch to Intel at all. The option to be ABLE to was certainly important, though, and was an option Apple had long been working on.)
If this contributed to the NeXT choice and the return of Steve, then it contributed to OS X, iPod, iOS, iCloud, every Apple product line sold today, AND therefore, the shape of the computer AND phone AND music industries today, as well as the entire touch-computing paradigm that has begun slowly replacing the mouse for everyday computing! It made Android possible (as more than the Blackberry clone it was pre-iPhone). Even Windows Phone 7: it?s not an iPhone-mimic like Android, but without iPhone, and without iPhone causing the rise of Android, it?s likely Microsoft would never have developed Windows Phone 7 as we know it.
I really think there is more to it than 'we needed an OS that could run on Intel'. Apple spent years trying to develop a modern OS, to replace the original MacOS and failed! At that point, their ONLY option was to acquire an OS and put their UI on it. In addition to NeXT and Be, Apple was looking at other Unix variants such as Sun Solaris which ran on SPARC, Intel, and PowerPC. They ultimately chose NeXT, but other than the fact that NeXTStep (renamed 'OpenStep') was being de-emphasized in favor of NetObjects, Tesler's comments do NOT tell the whole story...
Yaay! Perspective (and not these posts only).
.tsooJ
Did they discuss why Rosetta was retired?
I suspect that Apple has a new Rosetta that allows x86 compiled app to run on ARM
Copland was a project at Apple Computer to create an updated version of the Macintosh operating system. It was to have introduced protected memory, preemptive multitasking and a number of new underlying operating system features, yet still be compatible with existing Mac software. ...
which coincides with what I recall actually going on at the time.
The early '90s was the period when preemptive multitasking and a number of other features we now take for granted were introduced to personal computer operating systems. Copland's stillbirth was a major problem for Apple, and, at least in retrospect, symptomatic of much bigger problems at the company.
I think this story about interest in an Intel switch being the reason for acquiring NeXT is a rewriting of history. Apple desperately needed a modern operating system and the fact that NeXTSTEP was already running on Intel was just a bonus for them, one that wasn't squandered.
Did they discuss why Rosetta was retired?
Rosetta was the second emulator that Apple had to implement. The first being the 68K emulator that was part of Classic.
I believe that Rosetta involved a considerable number of hooks in OS X to make it work correctly. EOL'ing Rosetta cleaned up the build tree I'm sure. At this point they knew the remaining number of non-UB apps was minimal, and there were sufficient replacements for the few that never transitioned.
Pity they never re-compiled AppleWorks 6 as a UB. I wonder if this was a business decision (in light of iWorks) or if AW6 had sufficient non-portable code down deep within. I would care a lot less about this one, had Apple implemented a proper database module as part of iWorks (ignoring Bento for this discussion). Anyone with legacy data in AW6 is stuck at 10.6.8.
BTW, this article finally adds color to where/when Apple soured on PPC. Up until now, most people thought it happened after the NeXT acquisition. Now that we see it goes back much farther, then it appears to be a matter of timing. The promised G5 at 3 Ghz, which IBM missed the deadline on, was probably the tipping point that pushed the decision over the edge.
I suspect that Apple has a new Rosetta that allows x86 compiled app to run on ARM
Rosetta only worked on i86 because PPC was such a terrible professor. I said so in the forums years ago and people ignored me. The fact I'd PPC processors had terrible integer performance while intel was many times better. This additional integer capacity allowed for impressive emulation of PPC code.
ARM is already at a deficit with respect to intel hardware so any sort of Intel emulation will be pretty pathetic. The only way ARM will be viable near term is if the machine is running native code or hot compiling code with the facilities of LLVM. If not native performance would more than suck.
I suspect that Apple has a new Rosetta that allows x86 compiled app to run on ARM
I doubt that. But an ARM/iOS based desktop (similar to the iMac) is more likely. That way, the switchers who are buying iOS devices at a fast clip can run the same app on a home based desktop. The gestures would have to be done via the Magic Trackpad tho.
Intel processors of that era were HOT! Apple did not switch to Intel until a decade later. One major reason was the Intel processors of the mid-2000s were a lot cooler than the PowerPCs of the era. The other major reason was that IBM refused to design cooler versions of the PowerPC. Among other things, this meant that a PowerBook G4 was hot and that the PowerBook G5 was impossible to build.
Don't forget that the G5 (the last few used in the Power Mac) had to be water cooled. That is the only Apple product ever released that needed that level of cooling.
Steve + NeXT + Apple = Good.
Steve + Cancer + Limited time on earth = sadness.
Steve + Apple + intense desire and drive = happy Apple Lovers.
Apple - Steve = The Unknown