3 ms·

Both the iPhone 7 and the next next iPad Pro will be significantly faster than the current Macbook (and Macbook Air) and that's probably also the case for the n

by mdemare 10y ago

Both the iPhone 7 and the next next iPad Pro will be significantly faster than the current Macbook (and Macbook Air) and that's probably also the case for the next iteration, if it sticks with Intel. That alone is reason to switch to ARM. Compiling to ARM will be relatively easy, but still emulation will be needed for many applications, and that might be harder than emulating PowerPC. I wonder where the performance ceiling is going to be for the Apple processor line. Looking at Intel, I'd have expected them to have hit it already, but given the huge increases in single threaded performance so far year on year, it's hard to believe that it'll peter down to <15% per year from now on. If performance is going to keep improving at the current rate, then the A11X will be significantly faster than Intel's 2017 Cannonlake mobile lineup. Switching away from Intel would give Macs another big leg-up relative to Windows laptops. Macs will never switch to iOS though. Processes on macOS have a freedom that's incomparable to that on iOS, and once the genie is out of the bottle, you can't put it back in again. But Apple will keep working to make iOS sufficient for increasingly more people, and Macs will keep getting more niche. Apple could even introduce a laptop-like form factor like the Surface Book with iOS on it (the iBook Pro?). Interesting times...

agumonkey 10y ago
Wouldn't be surprised to hear Fabrice Bellard received a letter from Apple Inc.
- tonyarkles 10y ago
  For those who are scratching their heads wondering who this is, he's the author of qemu. It can already do arm and x86 emulation very well. Also, if you want to see a list of crazy shit this guy has done, see: http://bellard.org/ http://bellard.org/
jhasse 10y ago
"Both the iPhone 7 and the next next iPad Pro will be significantly faster than the current Macbook (and Macbook Air)" Source? I don't believe that this is true.
- jsmith0295 10y ago
  If you look up Geekbench scores for these devices at least, the iPhone 7 gets a higher score than the 12" MacBook, the latest MacBook Air, is similar in performance to a 2013 MacBook Pro, and even has better single core performance than the top of the line 12 core Mac Pro.
  - whamlastxmas 10y ago
    Read the HN comments on that article. It's bullshit.
    - kyrra 10y ago
      Which discussion, this one? It doesnt seem that clear cut from the comments. https://news.ycombinator.com/item?id=12511779 https://news.ycombinator.com/item?id=12511779
      - seanp2k2 10y ago
        Still apples to oranges, and it's just one benchmark. How long does e.g. Npm install take with the same project, compiling the same dependencies? ARM might come to desktop eventually, and it's damn impressive that for a single benchmark from a single test, someone was able to get the ip7 to beat a >1000 days-old Mac Pro in one metric, but they've learned from the PowerPC switch and a new arch won't happen overnight. I very highly doubt that the MacBooks hopefully coming later in 2016 will be anything other than Intel, hopefully Kaby Lake. I'd predict that the earliest we'd see an ARM MacBook would be 2018, and my guess is that they'd do it as a separate mode (so, have the A12X CPU in there with an Intel Coffee Lake or Cannon Lake main x86 CPU) and let it do something like dual-boot iOS and macOS, or have way tighter integration (run iOS apps as just another window in macOS, for example...dual-booting seems un-Apple-y, even with BootCamp).
  - jhasse 10y ago
    MacBook Air (11-inch Early 2015) 3236 iPhone 7 3288 I wouldn't call this "significantly", especially when looking at the multi-core scores: MacBook Air (11-inch Early 2015) 6419 iPhone 7 5308 The iPhone 7 was just released, the MacBook Air is over a year old.
valine 10y ago
I doubt we will ever see a laptop running iOS. iOS was designed solely for a touch interface. At the same time, apple has been outspoken about not wanting touch screens on laptops because of ergonomic issues and other things. http://www.theverge.com/2015/11/15/9738504/tim-cook-says-apple-wont-combine-a-macbook-and-ipad http://www.theverge.com/2015/11/15/9738504/tim-cook-says-app...
- AndrewKemendo 10y ago
  Well considering my surfacebook pro is a touch laptop/superbook I'd say it's a possibility.
- Avshalom 10y ago
  They will however happily sell you a $170 keyboard for your ipad... But snark aside, apple is always outspoken against things... until they release their own version of it. That's not some apple specific trait of course.
- on_and_off 10y ago
  They were also outspoken about how 3.5" was the perfect form factor.
andr 10y ago
How plausible is it to stick an ARM chip and a (lower-end) Intel CPU on the same board and make them work together? The transition to an ArmBook would be much smoother and I'm betting it won't affect cost of the laptop that much. After all, an Apple TV, which is essentially an iPhone 6 board without the battery and screen, costs just $149.
- api 10y ago
  That actually was done toward the tail end of the 1980s PC wars. The Commodore Amiga and several other machines could sport x86 add on boards that would let them run DOS apps natively. It would not be good for a laptop though, and that is almost all of Mac sales. Too much space and power consumption.
- wallacoloo 10y ago
  Pretty much every iteration of the Gameboy does this (at least up through the Nintendo DS) in order to achieve backward compatibility, so it has at least some plausibility.
jackmott 10y ago
faster by what measure? is claim supported by data?
sjtgraham 10y ago
Some thoughts on a X86 -> ARM migration: - Most 3rd party Apple development is already happening on ARM. This would inject new life into the desktop/laptop product line. Think of the thousands of apps that could now run on those platforms. - Apple gets rid of a massive chunk of BOM cost improving margins, i.e. Intel CPU. - No need to emulate X86 if Apple can do binary translation instead. There are already OSS projects like MC-Semantics (https://github.com/trailofbits/mcsema https://github.com/trailofbits/mcsema) that can take an X86 binary and emit LLVM IR. In theory you could give that to the ARM LLVM backend and relink to ARM frameworks. This would be an interesting problem to solve vis-a-vis code signing. One way would be for Apple to sign a code signing identity for each user, have the OS translate the binary, prompt the user for a password to get the private key from the keychain, sign the new executable, save it, and then run it. - I actually wonder how many desktop apps are commonly used on the macOS platform, and how big of a problem in 2016 an arch switch would be. Muggles are already predominantly using mobile devices. I'm guessing 99% of users use one of 100 (or less) desktop apps. I imagine it to be quite the long-tail.

by mdemare 10y ago

agumonkey 10y ago

Wouldn't be surprised to hear Fabrice Bellard received a letter from Apple Inc.

tonyarkles 10y ago
For those who are scratching their heads wondering who this is, he's the author of qemu. It can already do arm and x86 emulation very well. Also, if you want to see a list of crazy shit this guy has done, see: http://bellard.org/ http://bellard.org/

jhasse 10y ago

"Both the iPhone 7 and the next next iPad Pro will be significantly faster than the current Macbook (and Macbook Air)" Source? I don't believe that this is true.

jsmith0295 10y ago
If you look up Geekbench scores for these devices at least, the iPhone 7 gets a higher score than the 12" MacBook, the latest MacBook Air, is similar in performance to a 2013 MacBook Pro, and even has better single core performance than the top of the line 12 core Mac Pro.
- whamlastxmas 10y ago
  Read the HN comments on that article. It's bullshit.
  - kyrra 10y ago
    Which discussion, this one? It doesnt seem that clear cut from the comments. https://news.ycombinator.com/item?id=12511779 https://news.ycombinator.com/item?id=12511779
    - seanp2k2 10y ago
      Still apples to oranges, and it's just one benchmark. How long does e.g. Npm install take with the same project, compiling the same dependencies? ARM might come to desktop eventually, and it's damn impressive that for a single benchmark from a single test, someone was able to get the ip7 to beat a >1000 days-old Mac Pro in one metric, but they've learned from the PowerPC switch and a new arch won't happen overnight. I very highly doubt that the MacBooks hopefully coming later in 2016 will be anything other than Intel, hopefully Kaby Lake. I'd predict that the earliest we'd see an ARM MacBook would be 2018, and my guess is that they'd do it as a separate mode (so, have the A12X CPU in there with an Intel Coffee Lake or Cannon Lake main x86 CPU) and let it do something like dual-boot iOS and macOS, or have way tighter integration (run iOS apps as just another window in macOS, for example...dual-booting seems un-Apple-y, even with BootCamp).
- jhasse 10y ago
  MacBook Air (11-inch Early 2015) 3236 iPhone 7 3288 I wouldn't call this "significantly", especially when looking at the multi-core scores: MacBook Air (11-inch Early 2015) 6419 iPhone 7 5308 The iPhone 7 was just released, the MacBook Air is over a year old.

valine 10y ago

I doubt we will ever see a laptop running iOS. iOS was designed solely for a touch interface. At the same time, apple has been outspoken about not wanting touch screens on laptops because of ergonomic issues and other things. http://www.theverge.com/2015/11/15/9738504/tim-cook-says-apple-wont-combine-a-macbook-and-ipad http://www.theverge.com/2015/11/15/9738504/tim-cook-says-app...

AndrewKemendo 10y ago
Well considering my surfacebook pro is a touch laptop/superbook I'd say it's a possibility.
Avshalom 10y ago
They will however happily sell you a $170 keyboard for your ipad... But snark aside, apple is always outspoken against things... until they release their own version of it. That's not some apple specific trait of course.
on_and_off 10y ago
They were also outspoken about how 3.5" was the perfect form factor.

andr 10y ago

How plausible is it to stick an ARM chip and a (lower-end) Intel CPU on the same board and make them work together? The transition to an ArmBook would be much smoother and I'm betting it won't affect cost of the laptop that much. After all, an Apple TV, which is essentially an iPhone 6 board without the battery and screen, costs just $149.

api 10y ago
That actually was done toward the tail end of the 1980s PC wars. The Commodore Amiga and several other machines could sport x86 add on boards that would let them run DOS apps natively. It would not be good for a laptop though, and that is almost all of Mac sales. Too much space and power consumption.
wallacoloo 10y ago
Pretty much every iteration of the Gameboy does this (at least up through the Nintendo DS) in order to achieve backward compatibility, so it has at least some plausibility.

jackmott 10y ago

faster by what measure? is claim supported by data?

sjtgraham 10y ago

Some thoughts on a X86 -> ARM migration: - Most 3rd party Apple development is already happening on ARM. This would inject new life into the desktop/laptop product line. Think of the thousands of apps that could now run on those platforms. - Apple gets rid of a massive chunk of BOM cost improving margins, i.e. Intel CPU. - No need to emulate X86 if Apple can do binary translation instead. There are already OSS projects like MC-Semantics (https://github.com/trailofbits/mcsema https://github.com/trailofbits/mcsema) that can take an X86 binary and emit LLVM IR. In theory you could give that to the ARM LLVM backend and relink to ARM frameworks. This would be an interesting problem to solve vis-a-vis code signing. One way would be for Apple to sign a code signing identity for each user, have the OS translate the binary, prompt the user for a password to get the private key from the keychain, sign the new executable, save it, and then run it. - I actually wonder how many desktop apps are commonly used on the macOS platform, and how big of a problem in 2016 an arch switch would be. Muggles are already predominantly using mobile devices. I'm guessing 99% of users use one of 100 (or less) desktop apps. I imagine it to be quite the long-tail.