5 ms·

That particular "fact" does have some substance to it- In the past, Apple has switched as they feel that CPU horsepower is holding them back (keep in mind that

That particular "fact" does have some substance to it- In the past, Apple has switched as they feel that CPU horsepower is holding them back (keep in mind that they really want people to buy new every year) and they have already started moving their lineups over to ARM- their iPads have been ARM forever, and they have a pretty big knowledge base in it now. EDIT: I think you may have been referring to the 9to5mac article- I'm going to leave this comment here, but the current article is way more specific.

bshipp 7y ago
correct me if I'm wrong, but isn't the reason they moved to Intel chips in the first place because their homegrown CPUs couldn't keep up? Has that calculus changed in the meantime?
- Doctor_Fegg 7y ago
  PowerPC wasn’t really homegrown.
  - gumby 7y ago
    Though Apple was part of the original PowerPC consortium, and back then also had a chip team. Then again Apple was also the other partner (with Acorn) in founding ARM.
- esmi 7y ago
  They switched because Freescale and IBM couldn’t keep up with Intel. Apple did not make CPUs at that time. https://en.m.wikipedia.org/wiki/Apple%27s_transition_to_Intel_processors https://en.m.wikipedia.org/wiki/Apple%27s_transition_to_Inte...
  - bshipp 7y ago
    Fair enough, I always associated PowerPC with Apple and assumed they were the driving force behind it's use. Regardless, that makes the financial calculus even more suspect as they would have even fewer units to amortize the cost over. But I don't work for Apple, and quite frankly the CPU market could certain use the injection of some more competition.
    - monocasa 7y ago
      They helped a little with the overarching design (the ISA of the vector unit was pretty much Apple designed) but they never made any chips themselves AFAIK.
    - IOT_Apprentice 7y ago
      Fewer units? They have arm processors in every iPhone, iPad and AppleTV. At least 70 million units a quarter. The Mac CPUs would need more cores/threads (perhaps octocore).
    - protomyth 7y ago
      Apple and assumed they were the driving force behind it's use Quite the opposite. Apple, according to the book The Race for a New Game Machine, was an unprofitable demanding customer. The scale of the XBox 360 and PS3 dwarfed Apple's usage and IBM made several decisions that were contrary to Apple's needs. PowerPC for the embedded market has been the primary driver of PowerPC development.
      - rbanffy 7y ago
        At some point IBM offered Apple the Cell processor. I'm glad Apple opted to move to Intel then. The Cell is the hardest thing I ever tried to write code for.
        protomyth 7y ago
        IBM made decisions about the Cell PowerPC that made it unsuitable for Apple (no out of order) and offered the 970 as compensation. We all got to see how those promises played out.
- tyingq 7y ago
  I think a lot has changed. Maybe not the ability to consistently lead Intel, but at least a larger portion of end users that would find an ARM laptop "good enough" exists now.
  - noir_lord 7y ago
    AMD is beating intel on price and performance and they have a fraction of the resources of Apple. Intel is behind on process and has been for a couple of generations right as everyone hits a process wall. For the first time since I got into computers in the late 80s Intel looks vulnerable. As to whether Apple would go ARM on a MacBook who knows, I’d guess they wouldn’t but they could straight buy AMD in cash so it’s not beyond the realm they could tool up for desktop class ARM.
    - bshipp 7y ago
      Indeed, AMD is finally in a good position after years wandering the desert. That said, this current strong performance is the result of years of R&D, something that would have to be built from the ground up if Apple was to try and replicate it. I can't help but think there must be some secret benefit to make them pursue such a risky endeavour for something so tangential to their comparative advantage. Apple doesn't sell custom components, they sell consumer experiences. They must be planning on going a completely different direction with iOS to warrant custom processor technology.
      - sprafa 7y ago
        I suppose hiring people like this (and this is he one we hear about they could have hired 20-30 people without anyone noticing) is a way to shortcut such R&D
      - IOT_Apprentice 7y ago
        Umm, you do realize that Apple has a large design team and outsources Fab to TSMC? Their iPhone/iPAD/AppleTV hardware platforms already run on their own ARM chip design. They SHIP over 70 MILLION units a quarter. They have the R&D teams already churning out new designs each year. How is this risky?
monocasa 7y ago
The difference here is that the previous ISA changes have been accompanied with massive benefits. The 68k to ppc transition came with ppc being able to emulate 68k code faster than any 68k you could buy (in an interpreter, no JIT!!!), and the native ppc code was just ungodly fast comparitively. Ppc to x86 was less dramatic on raw perf, but was accompanied by a huge gain wrt perf per watt, and you could still emulate ppc on Intel about as well as you could run on a native ppc (with a JIT this time). But now, Moore's law is in the process of hitting a brick wall. Those big wins for free just aren't there. And if there's a x86 to arm JIT, it's either going to be single core, or the arm cores are going to have to have a stronger x86 like memory model, thus negating a huge part of their architectural win. IMO, Apple's going to make their own x86 chips because the base x86_64 patents are going to expire real soon now, and they have enough patents from the crazy amount of fabless acquisitions to potentially cross license the rest. Processors are commodities now, and that's really going to shake things up in crazy ways.
- mutt2016 7y ago
  Also Intel isn't the only maker anymore. Anybody can have a chip made by that other company TSMC and whatnot.
- NikkiA 7y ago
  > The 68k to ppc transition came with ppc being able to emulate 68k code faster than any 68k you could buy (in an interpreter, no JIT!!!) This simply isn't true, it wasn't until the G3-300 was released that the 68k emulator could run 68k code at 40MHz comparable speeds, and that was after 10 years of ppc601s and 603s running 68k system binaries at slower than quadra speeds on 180 and 200MHz processors.
- scarface74 7y ago
  The 68k to ppc transition came with ppc being able to emulate 68k code faster than any 68k you could buy (in an interpreter, no JIT!!!), and the native ppc code was just ungodly fast comparitively. Neither part of that statement is true. I had an LCII with a 68030-40Mhz card before upgrading to a 6100/60. The 6100/60 (601-60Mhz processor with a half speed bus) was much slower running 68K software than my LCII. It was about the same speed once I bought SpeedDoubler. Also, between the slower bus, the slow shared graphics memory and System 7 on the PPC being emulated, it felt much slower than a decent 486-DX2/66 or a Pentium-60.

bshipp 7y ago

correct me if I'm wrong, but isn't the reason they moved to Intel chips in the first place because their homegrown CPUs couldn't keep up? Has that calculus changed in the meantime?

Doctor_Fegg 7y ago
PowerPC wasn’t really homegrown.
- gumby 7y ago
  Though Apple was part of the original PowerPC consortium, and back then also had a chip team. Then again Apple was also the other partner (with Acorn) in founding ARM.
esmi 7y ago
They switched because Freescale and IBM couldn’t keep up with Intel. Apple did not make CPUs at that time. https://en.m.wikipedia.org/wiki/Apple%27s_transition_to_Intel_processors https://en.m.wikipedia.org/wiki/Apple%27s_transition_to_Inte...
- bshipp 7y ago
  Fair enough, I always associated PowerPC with Apple and assumed they were the driving force behind it's use. Regardless, that makes the financial calculus even more suspect as they would have even fewer units to amortize the cost over. But I don't work for Apple, and quite frankly the CPU market could certain use the injection of some more competition.
  - monocasa 7y ago
    They helped a little with the overarching design (the ISA of the vector unit was pretty much Apple designed) but they never made any chips themselves AFAIK.
  - IOT_Apprentice 7y ago
    Fewer units? They have arm processors in every iPhone, iPad and AppleTV. At least 70 million units a quarter. The Mac CPUs would need more cores/threads (perhaps octocore).
  - protomyth 7y ago
    Apple and assumed they were the driving force behind it's use Quite the opposite. Apple, according to the book The Race for a New Game Machine, was an unprofitable demanding customer. The scale of the XBox 360 and PS3 dwarfed Apple's usage and IBM made several decisions that were contrary to Apple's needs. PowerPC for the embedded market has been the primary driver of PowerPC development.
    - rbanffy 7y ago
      At some point IBM offered Apple the Cell processor. I'm glad Apple opted to move to Intel then. The Cell is the hardest thing I ever tried to write code for.
      - protomyth 7y ago
        IBM made decisions about the Cell PowerPC that made it unsuitable for Apple (no out of order) and offered the 970 as compensation. We all got to see how those promises played out.
tyingq 7y ago
I think a lot has changed. Maybe not the ability to consistently lead Intel, but at least a larger portion of end users that would find an ARM laptop "good enough" exists now.
- noir_lord 7y ago
  AMD is beating intel on price and performance and they have a fraction of the resources of Apple. Intel is behind on process and has been for a couple of generations right as everyone hits a process wall. For the first time since I got into computers in the late 80s Intel looks vulnerable. As to whether Apple would go ARM on a MacBook who knows, I’d guess they wouldn’t but they could straight buy AMD in cash so it’s not beyond the realm they could tool up for desktop class ARM.
  - bshipp 7y ago
    Indeed, AMD is finally in a good position after years wandering the desert. That said, this current strong performance is the result of years of R&D, something that would have to be built from the ground up if Apple was to try and replicate it. I can't help but think there must be some secret benefit to make them pursue such a risky endeavour for something so tangential to their comparative advantage. Apple doesn't sell custom components, they sell consumer experiences. They must be planning on going a completely different direction with iOS to warrant custom processor technology.
    - sprafa 7y ago
      I suppose hiring people like this (and this is he one we hear about they could have hired 20-30 people without anyone noticing) is a way to shortcut such R&D
    - IOT_Apprentice 7y ago
      Umm, you do realize that Apple has a large design team and outsources Fab to TSMC? Their iPhone/iPAD/AppleTV hardware platforms already run on their own ARM chip design. They SHIP over 70 MILLION units a quarter. They have the R&D teams already churning out new designs each year. How is this risky?

monocasa 7y ago

The difference here is that the previous ISA changes have been accompanied with massive benefits. The 68k to ppc transition came with ppc being able to emulate 68k code faster than any 68k you could buy (in an interpreter, no JIT!!!), and the native ppc code was just ungodly fast comparitively. Ppc to x86 was less dramatic on raw perf, but was accompanied by a huge gain wrt perf per watt, and you could still emulate ppc on Intel about as well as you could run on a native ppc (with a JIT this time). But now, Moore's law is in the process of hitting a brick wall. Those big wins for free just aren't there. And if there's a x86 to arm JIT, it's either going to be single core, or the arm cores are going to have to have a stronger x86 like memory model, thus negating a huge part of their architectural win. IMO, Apple's going to make their own x86 chips because the base x86_64 patents are going to expire real soon now, and they have enough patents from the crazy amount of fabless acquisitions to potentially cross license the rest. Processors are commodities now, and that's really going to shake things up in crazy ways.

mutt2016 7y ago
Also Intel isn't the only maker anymore. Anybody can have a chip made by that other company TSMC and whatnot.
NikkiA 7y ago
> The 68k to ppc transition came with ppc being able to emulate 68k code faster than any 68k you could buy (in an interpreter, no JIT!!!) This simply isn't true, it wasn't until the G3-300 was released that the 68k emulator could run 68k code at 40MHz comparable speeds, and that was after 10 years of ppc601s and 603s running 68k system binaries at slower than quadra speeds on 180 and 200MHz processors.
scarface74 7y ago
The 68k to ppc transition came with ppc being able to emulate 68k code faster than any 68k you could buy (in an interpreter, no JIT!!!), and the native ppc code was just ungodly fast comparitively. Neither part of that statement is true. I had an LCII with a 68030-40Mhz card before upgrading to a 6100/60. The 6100/60 (601-60Mhz processor with a half speed bus) was much slower running 68K software than my LCII. It was about the same speed once I bought SpeedDoubler. Also, between the slower bus, the slow shared graphics memory and System 7 on the PPC being emulated, it felt much slower than a decent 486-DX2/66 or a Pentium-60.