4 ms·

According to the article, Apple's CEO said they chose NeXT over Be because "In the end it came down to NeXT already supporting Intel and that was important to u

According to the article, Apple's CEO said they chose NeXT over Be because "In the end it came down to NeXT already supporting Intel and that was important to us". Why was running on Intel important to Apple? Apple didn't use Intel at that time (Dec 1996), and didn't decide to use Intel until much later (announced 2005)? Whereas BeOS already did run on the CPU platform that Apple did use at that time?

peatmoss 8y ago
My hunch is that they were absolutely hedging and that they liked not being dependent on a single CPU vendor.
cmiles74 8y ago
I purchased one of the first Macs that ran on a PowerPC processor and I was profoundly disappointed. Performance was no better than a 68040-based Mac and it was pretty crashy. As time went on and software was updated that machine became less crashy but performace was always poor. I don't have any insider information, but I wouldn't be surprised if Apple felt like they needed a backup plan in case the PowerPC alliance was unsuccessful at matching the speed growth of Intel's products. I'd say the G3 was the first PowerPC CPU that impressed me, I think Apple only stuck with PowerPC for two more generations (the G5).
- Tloewald 8y ago
  I got one of the first PowerPC computers in a lab full of high end spare no expense PCs and it cranked. If your experience was otherwise, I suspect you were running software under emulation so the fact it was “no better” was a borderline miracle. The second gen PowerPC macs were “the fastest 680x0 boxes ever shipped” (an Apple engineer boasted to me at WWDC and he wasn’t wrong) The PowerPC allowed Apple to overtake the Pentium and gave it boasting rights until the failure of the G5.
  - warrenm 8y ago
    And the G5 really only failed because of cooling issues - obviously, IBM has been able to continue to push the POWER platform that undergirded PowerPC far beyond the G5
    - bitwize 8y ago
      Not just cooling -- the G5 drew too much power to be viable in one of Apple's growth markets at the time: laptops. The Intel Core chips were based on the Pentium M microarchitecture and so already had power characteristics that were suited for laptop use. Of course, POWER would go on to be a great arch for desktop and server use, but that wasn't enough for Apple after the G4.
      - warrenm 8y ago
        Excessive power draw ... which, while seen in abysmal battery life, was most directly seen in an inability to cool
  - cmiles74 8y ago
    Yes, I believe the majority of software and even parts of the OS ran in emulation. Still, even when they moved a lot of that code native, performance wasn't amazing. Someone else mentioned that the rest of the hardware may have been holding these machines back and I could easily believe that. My opinion: when released, performance didn't meet my expectations and failed to meet my expectations for another five years (when the G3 was released). IMHO, I don't believe the Mac was competitive against the PC until the release of the G4. I could understand why Apple would want to hedge against the PowerPC.
    - bitwize 8y ago
      That's funny, I too remember things differently. As Mac advocates often, and loudly pointed out, the Mac consistently smoked the PC -- especially at the only benchmark that mattered, Photoshop filters. It was the 90s and RISC architecture really was gonna change everything. At the time, Intel was playing the "megahertz myth" heavily in its marketing, the false belief that more MHz made for a faster CPU, which just wasn't true when comparing across different microarchitectures like Pentium and PowerPC. And of course, as those from the Lost Amiga Civilization recall, it took about a 500 MHz Windows 9x PC to feel as responsive as a 25 MHz Amiga.
      - cmiles74 8y ago
        It's been a long time, my memory is less a collection of data and more a hazy soup of anecdotes. I do suspect that Photoshop filters might get some real performance gains out of AltiVec instructions, maybe my boring developer style workloads weren't reaping similar benefits from the availability of those instructions.
    - gowld 8y ago
      What the G4 giveth, Objective-C/Carbon/Cocoa taketh away.
- scarface74 8y ago
  The PPC processors were not the problem. The performance issues were caused by both a crippled bus on the first gen PPC computers and MacOS wasn’t fully native.
  - jdcarter 8y ago
    Right, BeOS ran like a bat out of hell even on low-end 603 PowerPC's. The 68K emulation in MacOS worked surprisingly well in terms of compatibility, but at a significant performance cost.
- mixmastamyk 8y ago
  Most of the software was emulated early on, so not completely unexpected.
- scruffyherder 8y ago
  The computer lab manager at the Uni I went to was obsessed with Apple, and he had one of the 601 based PowerMac's as well. And yes, I can confirm how underwhelming it was. Granted much of the OS was 68000 based, which certainly didn't help things in the slightest, but even Wolfenstein 3D on the thing really didn't run all that much better than our existing 486DX/2 66's. Running on Intel certainly saved NeXT in that by version 3.3 with the ability to support IDE CD-ROM's the cost of a machine that could run NeXTSTEP 3.3 had dropped dramatically. It's just a shame that the OS cost nearly as much as the machine. Being more open like the later Darwin sure may have helped, although I don't think NeXT much like later on Apple really wanted to either take in code from outsiders, or made the open software rediably usable.
dwighttk 8y ago
How long would it have taken to get BeOS on Intel to OS X 10.5 level? Or, perhaps Apple was considering switching earlier than they did. Maybe they thought G3->G4 was going to experience the problems that didn't end up cropping up until G4->G5. Apple was working on X86 compatibility as early as Mac OS 7 in 1992. [1] [1]https://en.wikipedia.org/wiki/Star_Trek_project https://en.wikipedia.org/wiki/Star_Trek_project
- scruffyherder 8y ago
  I don't think it'd matter for the future of Apple though. Since Steve had purchased Pixar, he had that 'in' for the entertainment industry and he was able to squeeze both music and film to get content for his iPod. Once they put the iPod on Windows Apple became a far more profitable media company. The iPhone just further cemented Apple's position as the device people want to carry to consume media. I don't see how going with Gasse or Be would have made the Apple of today possible.
S_A_P 8y ago
I think they knew at the time that Motorola processors weren’t quite living up to their promise. Theory didn’t seem to line up with practice and while they were technically faster per clock it didn’t always translate to performance. Intel also cleared some pretty incredible technical hurdles during this time period as well. I’ve no doubt that internally Apple engineers knew this.
Tloewald 8y ago
Maybe it was really because NeXTStep could print.
damon_c 8y ago
It's because back then, if I remember correctly, new revisions of Macintoshes with PowerPC processors had CPU speed increments like, 80, 90, 100mhz, and meanwhile the Pentium was going 100, 200, 300, 400... (at least that's how it felt being a mac user...) Also, I believe around that time the deal between IBM and Motorola was ending and they had already decided not to continue development of the PowerPC.
- mixmastamyk 8y ago
  Probably a factor. The money coming into Intel from the PC biz probably dwarfed Motorola's, leading to much bigger R&D budgets.
- spamizbad 8y ago
  In the G4 era, the marketing back then from Apple talked about the "Mhz Myth" and largely showcased stuff that was pretty much "Hey, this hand-tuned AltiVec code blows away this C++ -> x87 code on intel!" It's an interesting bit of history about how many very bright engineers had zero faith in intel being able pull off what they did with the P6 micro architecture: A superscalar, out-of-order x86 CPU that supported SMP on a power budget that, while high for the time (25 watts!), wasn't astronomical. That breakthrough ultimately became the foundation for Intel's Core micro architecture, which Apple later adopted in their first intel-based machines. If Apple had switched to Intel with the arrival of the Pentium, things could have been very different.
- Jetrel 8y ago
  Macs basically kept pace, pretty nose-to-nose, with the Mhz speed of Intel CPUs until the Pentium II era. Apple actually beat them to about 350mhz on the 604e for a year or so, whilst the fastest Pentium Pro of the time was about 200mhz. This was quite a stretch of time (92->97?). What happened after that, though, was that Intel hit the afterburners, and rocketed up to about 800mhz, and PPC got left in the dust. There was a huge 500mhz speed bump that the G3 and G4 both stalled at for a few years, and by the time they got over that, the pentium 4 was coming out, which started another jump from the P3's 1.5ghz, pushing the P4 up to about 3.5ghz. The G3 and G4 were much faster than either the 603 or P2/P3 and especially the P4, per mhz, so the launch day ad campaign saying the iMac's 233mhz CPU was "twice as fast" as the 250mhz pentiums that were out at the same time was true - it just didn't stay true once 500 mhz P2 machines came out less than a year later. It's notable that intel was so focused on the marketability of that metric that the later pentium series were worse per clock than their predecessors - a P4 had to actually be several hundred mhz faster than a P3 to be the same speed. Intel was primarily doing this to fight AMD, and it got bad enough that AMD actually had to introduce a new marketing term to indicate how fast their chips were compared to the mhz ratings of P4s. Intel completely reversed course on this with the core series - modern "core" chips like the i5 and i7 are several times faster than P4s running at the same clock speed. Far better heat/power consumption specs, too.
warrenm 8y ago
Remember how when Steve announced the transition to x86, he mentioned they'd already been working on it for a few years?
- scruffyherder 8y ago
  It goes all the way back to the release of NeXTSTEP 3.1, which included i386 support. And more importantly it never ended. The Rhapsody DR1/DR2 releases that would shape up to become OS X Server 1.0 were intel based. And the then released source code for Darwin 0.1/Darwin 0.3 although commercially only released on the PowerPC, still compiles and runs on i386 just fine. I'm sure with the overhaul of the kernel while going from OS X Server 1.0 to OS X 10.0 where the Objective C driver model was dumped for a C++ framework, along with updating the Mach code from 2.5 to 3.0 broke the i386, but it was quickly fixed in the Darwin releases, making the overall chain unbroken from the inside. It wouldn't surprise me in the lost years of Apple looking for a plan out of the demise of the 68000 that there would have been projects looking again and again at Intel, especially after the Star Trek project.
empath75 8y ago
It's possible they might have considered giving up on hardware and making a pc-compatible macos?
- dfox 8y ago
  In early 00's Apple actually provided x86 binaries for the Darwin (in form of essentially complete rpm-based distribution with Darwin kernel). Obviosly this didn't include the high-level NeXT/MacOS X ObjectiveC frameworks.
  - lukeh 8y ago
    And Rhapsody (Mac OS X Server 1.0) shipped a developer preview on Intel, including UI (this was pre-Aqua and Mach 3.0, but otherwise a lot of similarities to MacOS X 10.0).
  - scruffyherder 8y ago
    As a shameless plug, the Darwin 0.1/0.3 sources which align with OS X Server 1.0 & 1.2 actually still compiled on intel just fine. Even better you could take a Rhapsody DR2 system, and overlay the newer stuff on top of it, and it'd still happily run. https://virtuallyfun.com/wordpress/2017/04/22/darwin-0-1-rhapsody-dr-2-booted/ https://virtuallyfun.com/wordpress/2017/04/22/darwin-0-1-rha...
- adrianmsmith 8y ago
  That's a good theory, and there's precedent for it. Apple already sold their OS to Mac clone makers, and NeXT (earlier) and Be (later) did indeed gave up hardware.
jldugger 8y ago
Honestly, it sounds like an answer you give to avoid disclosing the real reason: relationships and cultural compatibility.
patrickg_zill 8y ago
NeXT was far, far more mature as an OS than BeOS. For years, people had been shipping mission-critical custom software on NeXT. You had Improv, which when Lotus killed it, was cloned into Quantrix which is still sold today to financial modelers. You had the McCaw Cellular custom app (don't know what it was called) that allowed them to grow - they later became Cingular, then ATT Wireless. William Morris Agency, Booz Allen Hamilton (big consulting firm), etc. all had custom apps that they used under NeXT. In contrast, BeOS had a few small desktop apps and some other cool demos.
kalleboo 8y ago
At the time Apple was licensing MacOS to clone manufacturers like Power Computing. It's likely they thought the only way to grow the clone business was to open the floodgates to Intel PC clone manufacturers. Of course the NeXT merger brought back Steve Jobs who killed the clone program immediately.

peatmoss 8y ago

My hunch is that they were absolutely hedging and that they liked not being dependent on a single CPU vendor.

cmiles74 8y ago

I purchased one of the first Macs that ran on a PowerPC processor and I was profoundly disappointed. Performance was no better than a 68040-based Mac and it was pretty crashy. As time went on and software was updated that machine became less crashy but performace was always poor. I don't have any insider information, but I wouldn't be surprised if Apple felt like they needed a backup plan in case the PowerPC alliance was unsuccessful at matching the speed growth of Intel's products. I'd say the G3 was the first PowerPC CPU that impressed me, I think Apple only stuck with PowerPC for two more generations (the G5).

Tloewald 8y ago
I got one of the first PowerPC computers in a lab full of high end spare no expense PCs and it cranked. If your experience was otherwise, I suspect you were running software under emulation so the fact it was “no better” was a borderline miracle. The second gen PowerPC macs were “the fastest 680x0 boxes ever shipped” (an Apple engineer boasted to me at WWDC and he wasn’t wrong) The PowerPC allowed Apple to overtake the Pentium and gave it boasting rights until the failure of the G5.
- warrenm 8y ago
  And the G5 really only failed because of cooling issues - obviously, IBM has been able to continue to push the POWER platform that undergirded PowerPC far beyond the G5
  - bitwize 8y ago
    Not just cooling -- the G5 drew too much power to be viable in one of Apple's growth markets at the time: laptops. The Intel Core chips were based on the Pentium M microarchitecture and so already had power characteristics that were suited for laptop use. Of course, POWER would go on to be a great arch for desktop and server use, but that wasn't enough for Apple after the G4.
    - warrenm 8y ago
      Excessive power draw ... which, while seen in abysmal battery life, was most directly seen in an inability to cool
- cmiles74 8y ago
  Yes, I believe the majority of software and even parts of the OS ran in emulation. Still, even when they moved a lot of that code native, performance wasn't amazing. Someone else mentioned that the rest of the hardware may have been holding these machines back and I could easily believe that. My opinion: when released, performance didn't meet my expectations and failed to meet my expectations for another five years (when the G3 was released). IMHO, I don't believe the Mac was competitive against the PC until the release of the G4. I could understand why Apple would want to hedge against the PowerPC.
  - bitwize 8y ago
    That's funny, I too remember things differently. As Mac advocates often, and loudly pointed out, the Mac consistently smoked the PC -- especially at the only benchmark that mattered, Photoshop filters. It was the 90s and RISC architecture really was gonna change everything. At the time, Intel was playing the "megahertz myth" heavily in its marketing, the false belief that more MHz made for a faster CPU, which just wasn't true when comparing across different microarchitectures like Pentium and PowerPC. And of course, as those from the Lost Amiga Civilization recall, it took about a 500 MHz Windows 9x PC to feel as responsive as a 25 MHz Amiga.
    - cmiles74 8y ago
      It's been a long time, my memory is less a collection of data and more a hazy soup of anecdotes. I do suspect that Photoshop filters might get some real performance gains out of AltiVec instructions, maybe my boring developer style workloads weren't reaping similar benefits from the availability of those instructions.
  - gowld 8y ago
    What the G4 giveth, Objective-C/Carbon/Cocoa taketh away.
scarface74 8y ago
The PPC processors were not the problem. The performance issues were caused by both a crippled bus on the first gen PPC computers and MacOS wasn’t fully native.
- jdcarter 8y ago
  Right, BeOS ran like a bat out of hell even on low-end 603 PowerPC's. The 68K emulation in MacOS worked surprisingly well in terms of compatibility, but at a significant performance cost.
mixmastamyk 8y ago
Most of the software was emulated early on, so not completely unexpected.
scruffyherder 8y ago
The computer lab manager at the Uni I went to was obsessed with Apple, and he had one of the 601 based PowerMac's as well. And yes, I can confirm how underwhelming it was. Granted much of the OS was 68000 based, which certainly didn't help things in the slightest, but even Wolfenstein 3D on the thing really didn't run all that much better than our existing 486DX/2 66's. Running on Intel certainly saved NeXT in that by version 3.3 with the ability to support IDE CD-ROM's the cost of a machine that could run NeXTSTEP 3.3 had dropped dramatically. It's just a shame that the OS cost nearly as much as the machine. Being more open like the later Darwin sure may have helped, although I don't think NeXT much like later on Apple really wanted to either take in code from outsiders, or made the open software rediably usable.

dwighttk 8y ago

How long would it have taken to get BeOS on Intel to OS X 10.5 level? Or, perhaps Apple was considering switching earlier than they did. Maybe they thought G3->G4 was going to experience the problems that didn't end up cropping up until G4->G5. Apple was working on X86 compatibility as early as Mac OS 7 in 1992. [1] [1]https://en.wikipedia.org/wiki/Star_Trek_project https://en.wikipedia.org/wiki/Star_Trek_project

scruffyherder 8y ago
I don't think it'd matter for the future of Apple though. Since Steve had purchased Pixar, he had that 'in' for the entertainment industry and he was able to squeeze both music and film to get content for his iPod. Once they put the iPod on Windows Apple became a far more profitable media company. The iPhone just further cemented Apple's position as the device people want to carry to consume media. I don't see how going with Gasse or Be would have made the Apple of today possible.

S_A_P 8y ago

I think they knew at the time that Motorola processors weren’t quite living up to their promise. Theory didn’t seem to line up with practice and while they were technically faster per clock it didn’t always translate to performance. Intel also cleared some pretty incredible technical hurdles during this time period as well. I’ve no doubt that internally Apple engineers knew this.

Tloewald 8y ago

Maybe it was really because NeXTStep could print.

damon_c 8y ago

It's because back then, if I remember correctly, new revisions of Macintoshes with PowerPC processors had CPU speed increments like, 80, 90, 100mhz, and meanwhile the Pentium was going 100, 200, 300, 400... (at least that's how it felt being a mac user...) Also, I believe around that time the deal between IBM and Motorola was ending and they had already decided not to continue development of the PowerPC.

mixmastamyk 8y ago
Probably a factor. The money coming into Intel from the PC biz probably dwarfed Motorola's, leading to much bigger R&D budgets.
spamizbad 8y ago
In the G4 era, the marketing back then from Apple talked about the "Mhz Myth" and largely showcased stuff that was pretty much "Hey, this hand-tuned AltiVec code blows away this C++ -> x87 code on intel!" It's an interesting bit of history about how many very bright engineers had zero faith in intel being able pull off what they did with the P6 micro architecture: A superscalar, out-of-order x86 CPU that supported SMP on a power budget that, while high for the time (25 watts!), wasn't astronomical. That breakthrough ultimately became the foundation for Intel's Core micro architecture, which Apple later adopted in their first intel-based machines. If Apple had switched to Intel with the arrival of the Pentium, things could have been very different.
Jetrel 8y ago
Macs basically kept pace, pretty nose-to-nose, with the Mhz speed of Intel CPUs until the Pentium II era. Apple actually beat them to about 350mhz on the 604e for a year or so, whilst the fastest Pentium Pro of the time was about 200mhz. This was quite a stretch of time (92->97?). What happened after that, though, was that Intel hit the afterburners, and rocketed up to about 800mhz, and PPC got left in the dust. There was a huge 500mhz speed bump that the G3 and G4 both stalled at for a few years, and by the time they got over that, the pentium 4 was coming out, which started another jump from the P3's 1.5ghz, pushing the P4 up to about 3.5ghz. The G3 and G4 were much faster than either the 603 or P2/P3 and especially the P4, per mhz, so the launch day ad campaign saying the iMac's 233mhz CPU was "twice as fast" as the 250mhz pentiums that were out at the same time was true - it just didn't stay true once 500 mhz P2 machines came out less than a year later. It's notable that intel was so focused on the marketability of that metric that the later pentium series were worse per clock than their predecessors - a P4 had to actually be several hundred mhz faster than a P3 to be the same speed. Intel was primarily doing this to fight AMD, and it got bad enough that AMD actually had to introduce a new marketing term to indicate how fast their chips were compared to the mhz ratings of P4s. Intel completely reversed course on this with the core series - modern "core" chips like the i5 and i7 are several times faster than P4s running at the same clock speed. Far better heat/power consumption specs, too.

warrenm 8y ago

Remember how when Steve announced the transition to x86, he mentioned they'd already been working on it for a few years?

scruffyherder 8y ago
It goes all the way back to the release of NeXTSTEP 3.1, which included i386 support. And more importantly it never ended. The Rhapsody DR1/DR2 releases that would shape up to become OS X Server 1.0 were intel based. And the then released source code for Darwin 0.1/Darwin 0.3 although commercially only released on the PowerPC, still compiles and runs on i386 just fine. I'm sure with the overhaul of the kernel while going from OS X Server 1.0 to OS X 10.0 where the Objective C driver model was dumped for a C++ framework, along with updating the Mach code from 2.5 to 3.0 broke the i386, but it was quickly fixed in the Darwin releases, making the overall chain unbroken from the inside. It wouldn't surprise me in the lost years of Apple looking for a plan out of the demise of the 68000 that there would have been projects looking again and again at Intel, especially after the Star Trek project.

empath75 8y ago

It's possible they might have considered giving up on hardware and making a pc-compatible macos?

dfox 8y ago
In early 00's Apple actually provided x86 binaries for the Darwin (in form of essentially complete rpm-based distribution with Darwin kernel). Obviosly this didn't include the high-level NeXT/MacOS X ObjectiveC frameworks.
- lukeh 8y ago
  And Rhapsody (Mac OS X Server 1.0) shipped a developer preview on Intel, including UI (this was pre-Aqua and Mach 3.0, but otherwise a lot of similarities to MacOS X 10.0).
- scruffyherder 8y ago
  As a shameless plug, the Darwin 0.1/0.3 sources which align with OS X Server 1.0 & 1.2 actually still compiled on intel just fine. Even better you could take a Rhapsody DR2 system, and overlay the newer stuff on top of it, and it'd still happily run. https://virtuallyfun.com/wordpress/2017/04/22/darwin-0-1-rhapsody-dr-2-booted/ https://virtuallyfun.com/wordpress/2017/04/22/darwin-0-1-rha...
adrianmsmith 8y ago
That's a good theory, and there's precedent for it. Apple already sold their OS to Mac clone makers, and NeXT (earlier) and Be (later) did indeed gave up hardware.

jldugger 8y ago

Honestly, it sounds like an answer you give to avoid disclosing the real reason: relationships and cultural compatibility.

patrickg_zill 8y ago

NeXT was far, far more mature as an OS than BeOS. For years, people had been shipping mission-critical custom software on NeXT. You had Improv, which when Lotus killed it, was cloned into Quantrix which is still sold today to financial modelers. You had the McCaw Cellular custom app (don't know what it was called) that allowed them to grow - they later became Cingular, then ATT Wireless. William Morris Agency, Booz Allen Hamilton (big consulting firm), etc. all had custom apps that they used under NeXT. In contrast, BeOS had a few small desktop apps and some other cool demos.

kalleboo 8y ago

At the time Apple was licensing MacOS to clone manufacturers like Power Computing. It's likely they thought the only way to grow the clone business was to open the floodgates to Intel PC clone manufacturers. Of course the NeXT merger brought back Steve Jobs who killed the clone program immediately.