ARM based MacBook Pro discussion [merged]

[pshufd]

Porting is usually a technical challenge - it's not that it can't be done. It's that you need time, people, tools, resources, and communications.

I didn't work on the JIT - I just sat in on some of the meetings. It would have been fun to work on but I had to care for a relative with severe health issues for the next three years and that soaked up a lot of my time.

These are my old x64 Windows builds. They started in 2005. I think that Mozilla started official nightly ports to x64 around 2011 or 2012. There are some old PPC G5 builds in there too - I was doing some Altivec experiments back then.

http://www.vector64.com/WindowsBuilds.html

> The ARM Windows version of Firefox is in hand,

That's a substantial achievement. Mozilla's code was designed to be portable as they probably started out on old Unix hardware and then had to run on Windows and other platforms. Doing a new platform is a slog of a job. Some things like datatypes aren't that bad. You implement the types for your architecture and operating system and then try a test compile and see what breaks and then you go around fixing it. Then you do assembler/machine code stuff, and hope that the intrinsics work. And once you get a good compile/link, you look at the crashes.

 

[kis]

Apples last event had one large bit of news that most people didn’t pick up on, ARM chips coming to Macs soon. The new service that will play games in iOS, TVOS and OSX, Apple Arcade is the proof. The service is launching Fall 2019, right in time for newest Macs with ARM chips to be released. In order to play the same games as IOS/TVOS chances are all new MACs will have ARM chips on board. Don’t forget, these will be IOS/TVOS (both currently have ARM chips) games ported and playable locally on MACs. It makes perfect sense to launch it along side the new ARM MACs.

They’re likely coming for hybrid devices like a laptop-version if the iPad Pro etc. They’re nowhere near ready for desktop computers and professional laptops. Sure, someday they will, but we’re looking at years here not months.

 

[kis]

You mean like the “full version” of Photoshop already demoed and slated for release on IOS?

You mean the “full” version that has proven to be so difficult to port that the development is almost a year behind?

 

[danwells]

I agree with the posters who are saying that we aren't far from saying an ARM Mac, but it'll start out with something like a 12" MacBook. There's an easy way for Apple to support both ARM and Intel (/Ryzen?) Macs simultaneously... Restrict ARM Macs to the Mac App Store. If ARM Macs (ultraportables, some low-end/no active cooling Minis, and possibly some iMacs) are Mac App Store only, Apple can very easily require all apps in the Mac App Store to provide either dual binaries or fat binaries. Meanwhile, the Intel Macs continue on running Mac App Store software, independently acquired software and Windows/Unix software through Parallels/VMWare/Boot Camp.

If I were Apple, I'd use the word Pro to differentiate Intel Macs? MacBook Pro stays Intel, while MacBook goes ARM... The powerful 27" iMac becomes a member of the iMac Pro line (while the Xeon iMac Pro becomes a 32" machine?) while the 21" iMac goes ARM, the Mac Pro of course stays Intel. There might be a Mac Mini Pro (Intel) that encompasses the higher end of today's Mini line which are often used as servers and compute boxes, while an ARM Mac Mini takes over the media-centric duties at the lower end.

 

[mcnallym]

Personally would be happy with an ARM mini.

mini now just used as an iTunes server, headless with a fit 4k hdmi adaptor attached to fool the system into thinking monitor attached. So ARM would be perfect for that. As long as something like iFlicks ported then would be happy.

Pretty sure that years ago ( whilst Steve still around ) that there were some ARM macs running in macbook air chassis. not for the public but internal use so wouldn't be surprised if there are ARM Macs already in the labs, running mac os.

My cMP is looking at replacing with either a new iMac 2019 with the i9 or an i9 hack, wait until the 10.14.5 out and maybe even get a Radeon VII for the heavy work. As long as can connect to an ARM mini with the Screen Sharing then wouldn't be fussed about ARM and Intel mix.

At the point now where as long as works together and fast enough for what I want I don't care what inside.

 

[collin_]

If Apple were to transition the MBP to its custom ARM CPUs, here are some potential (mostly overlapping) benefits we could see:

• The end of thermal issues. Possibly even MBPs that could handle sustained maximum loads. (Also consider that less heat is good for the long-term health of the machine... and our laps).
• Vastly more efficient Macs with much longer battery life (We would be going from 14nm chips to 7nm, or 5nm assuming they came out in late 2020. We would also likely see a mix of high performance and efficiency cores, as with Apple's A-series chips. Note that Macs would also probably age better due to the degradable batteries being cycled less quickly)
  
• Enough GPU performance to make Macs viable as gaming computers (I know it doesn't make sense to buy a Mac for this, but it would be a nice bonus. Consider that the 2018 iPad Pro already matches the graphics performance of an Xbox One S)
• More powerful Macs in general (The A12X processor already rivals or even exceeds the performance of the current Macbook Pros in many different areas. Imagine what Apple could do with more thermal headroom and a cooling system)
• Regular updates (Unlike Intel, TMSC has been consistently shrinking its node size and refining its manufacturing process)
• New features (Apple's custom silicon has enabled things like FaceID in iPhones and hardware acceleration for encryption in Macs. We would see more features like these over time)
• (Possibly) Cheaper Macs (The cost to manufacture each Mac, especially the ones with custom GPUs, would drop dramatically. Of course, it's unrealistic to think that Apple would pass on all the savings to the consumer, but they would likely pass on at least some of it. Tim Cook did say recently that he hopes to reduce the price of the Macbook Air in the future. I think this comment was hinting at Apple custom silicon)

Some have expressed concern that people who run certain specific programs (mainly virtual machines) could be in trouble. There would also obviously have to be some sort of macOS software to accommodate third-party applications, which would likely see slower performance for a while until they were updated to run on ARM (note: this sort of thing has already been done in Windows). It's impossible to say how much of a problem these things would be, because we don't yet know how Apple would deal with it. However, the applications everyone uses the most (e.g. Chrome) would probably update rapidly.

All that to say, I think the (massive) potential benefits of an ARM Macbook Pro far outweigh the potential drawbacks, and would be interested to hear your opinions.

 

[GoodGuy12345]

I am soo siked for this! Finally my laptop wont start to burn my lap and I would have to quickly find a table to set it on. A GPU performance bump would definitely be a good start as some people have Macs as their main computers, and cramming a bit of gaming on the side would be excellent.

 

[mdnz]

Good points, I'd like to see it on a lower end Macbook at the start to let all the kinks show up.

 

[Stephen.R]

No. If apple go back to non-Intel cpus across the board I’ll need to find something else to run Linux VMs on, which while not the end of the world at a desk, isn’t practical if I’m out somewhere with the laptop.

 

[Juicy Box]

I am not, at least for the Mac desktops, but I see the appeal for laptops.

 

[LogicalApex]

I'm not convinced we'll see this come to fruition honestly... Not unless computing changes dramatically first. Ditching Intel wouldn't be an easy choice for Apple as it isn't just about the CPU, but also about the package that Intel provides.

For instance, Thunderbolt 3 is an Intel specification that is core to the portability of the MacBook...

The last time Apple did an architecture swap the "Desktop" OS was the only option. With the entire industry moving mobile first if Apple were to force another transition they would effectively kill off their Mac business. ARM to X86 emulation isn't super performant either.... If it were you'd see tons of Windows laptops running ARM chips today and using Windows 10 on ARM which supports this emulation out of the box...

What we'll likely see is more Apple custom ARM chips powering enhancements to the Mac like the T2 is doing...

 

[pshufd]

Intel has various other technologies built into the CPU that you can read about on their CPU spec pages.

It's also not clear that ARM can run as well as Intel on a high-power process. And hardware-optimized code won't run well or at all.

 

[SoyCapitanSoyCapitan]

Some people think that ARM processors are some kind of toy CPUs for gadgets. They haven’t been around long enough to know ARM began as a desktop CPU and was used in some of the earliest visual workstations like the Acorn Archimedes. Those were faster than Macs at the time.

 

[pshufd]

Some people think that ARM processors are some kind of toy CPUs for gadgets. They haven’t been around long enough to know ARM began as a desktop CPU and was used in some of the earliest visual workstations like the Acorn Archimedes. Those were faster than Macs at the time.

Some people should ask chip designers and electrical engineers about the odds of ARM chips displacing Intel chips.

 

[jpn]

If Apple were to transition the MBP to its custom ARM CPUs, here are some potential (mostly overlapping) benefits we could see:

• The end of thermal issues. Possibly even MBPs that could handle sustained maximum loads. (Also consider that less heat is good for the long-term health of the machine... and our laps).
• Vastly more efficient Macs with much longer battery life (We would be going from 14nm chips to 7nm, or 5nm assuming they came out in late 2020. We would also likely see a mix of high performance and efficiency cores, as with Apple's A-series chips. Note that Macs would also probably age better due to the degradable batteries being cycled less quickly)
  
• Enough GPU performance to make Macs viable as gaming computers (I know it doesn't make sense to buy a Mac for this, but it would be a nice bonus. Consider that the 2018 iPad Pro already matches the graphics performance of an Xbox One S)
• More powerful Macs in general (The A12X processor already rivals or even exceeds the performance of the current Macbook Pros in many different areas. Imagine what Apple could do with more thermal headroom and a cooling system)
• Regular updates (Unlike Intel, TMSC has been consistently shrinking its node size and refining its manufacturing process)
• New features (Apple's custom silicon has enabled things like FaceID in iPhones and hardware acceleration for encryption in Macs. We would see more features like these over time)
• (Possibly) Cheaper Macs (The cost to manufacture each Mac, especially the ones with custom GPUs, would drop dramatically. Of course, it's unrealistic to think that Apple would pass on all the savings to the consumer, but they would likely pass on at least some of it. Tim Cook did say recently that he hopes to reduce the price of the Macbook Air in the future. I think this comment was hinting at Apple custom silicon)

Some have expressed concern that people who run certain specific programs (mainly virtual machines) could be in trouble. There would also obviously have to be some sort of macOS software to accommodate third-party applications, which would likely see slower performance for a while until they were updated to run on ARM (note: this sort of thing has already been done in Windows). It's impossible to say how much of a problem these things would be, because we don't yet know how Apple would deal with it. However, the applications everyone uses the most (e.g. Chrome) would probably update rapidly.

All that to say, I think the (massive) potential benefits of an ARM Macbook Pro far outweigh the potential drawbacks, and would be interested to hear your opinions.

hi

changing a platforms...
i willingly switched to OS X when it was released.
it was painful. but worth it. there were just a lot of the world that didn't change or shift quickly.
so the pain continued.
for a couple of years.

but, i am with you 100% on shifting to ARM for a MacBook Air level of laptop.
however, i don't see it being able to replace desktop or MacBook Pro level laptops.

apple has taken a an even longer strategic approach this time than it did when it shifted to Intel. which is great.
its making it easy for developers to write code that can be more easily modified to run on both Intel and ARM.
i would expect major announcements about this at WWDC in a few weeks.

i want a MacBook Air using ARM, yesterday, now even!
but actually when i say i want a ARMed Macbook Air, all that i am saying is that:
(1) apple only needs to continue to develop further the apps currently that work on an iOS envionment iPad so that the feature set is more powerful, and
(2) make iOS capable of handling a trackpad input on a Smart Keyboard.

then i have my ARMed Macbook Air level machine that I can use to actually get work done it!!!

the apps that apple has made available for use in macOS Mojave that try to mimic iOS apps found on an iPhone - like Stocks - are horrendous.
but there are also some apps that apple has made really good choices to try to blend iOS and macOS features, such as Pages, Numbers, and Keynote.
a lot a people would, however, quickly point out that apple has dumbed down these app's features during the past 3 years.

apple: trackpad support on Smart Keyboards. now please.

 

[poorcody]

I like the potential of it for all the reasons you list in the opening post. I am nervous, though, because running Windows in a VM is a must for me. If it works through some sort of emulation that will be fine (I don't need it to be blazing fast), but if that becomes impossible, I probably will have to give up using a MacBook...

 

[retta283]

I'm not. If devs have to recompile their Mac software for ARM, I can imagine many would just stop developing for the Mac. We'd be back to early-mid 2000s level software support. Can't have that. I think it would kill the Mac as a viable platform for many creatives.

 

[collin_]

I'm not convinced we'll see this come to fruition honestly... Not unless computing changes dramatically first. Ditching Intel wouldn't be an easy choice for Apple as it isn't just about the CPU, but also about the package that Intel provides.

For instance, Thunderbolt 3 is an Intel specification that is core to the portability of the MacBook...

The last time Apple did an architecture swap the "Desktop" OS was the only option. With the entire industry moving mobile first if Apple were to force another transition they would effectively kill off their Mac business. ARM to X86 emulation isn't super performant either.... If it were you'd see tons of Windows laptops running ARM chips today and using Windows 10 on ARM which supports this emulation out of the box...

What we'll likely see is more Apple custom ARM chips powering enhancements to the Mac like the T2 is doing...

Interesting points. Would it not be possible for ARM Macs to have Thunderbolt 3?

 

[poorcody]

Interesting points. Would it not be possible for ARM Macs to have Thunderbolt 3?

Yes, Intel opened up the license for Thunderbolt so it is now available for other chipmakers to use.

https://www.theverge.com/2019/3/4/18246182/usb-4-thunderbolt-3-specs-features-release-date

 

[cube]

I don't care as I don't expect to buy a Mac again.

 

[mavericks7913]

Intel CPU sucks and having tons of security issue. I dont mind if Apple makes ARM MacBook series. The only concern would be apps like Adobe.

Oh yes, no more thunderbolt support. TB3 is opened but that doeant mean TB4 can be open for all.

 

[pshufd]

One other aspect that most probably haven't experienced is the level of technical consulting that Intel provides to its customers.

 

[LogicalApex]

Yes, Intel opened up the license for Thunderbolt so it is now available for other chipmakers to use.

https://www.theverge.com/2019/3/4/18246182/usb-4-thunderbolt-3-specs-features-release-date

Not so fast... Intel still owns the patents for Thunderbolt 3 and and we don't have an idea of how they open their license is.

 

[pshufd]

Intel CPU sucks and having tons of security issue. I dont mind if Apple makes ARM MacBook series. The only concern would be apps like Adobe.

Oh yes, no more thunderbolt support. TB3 is opened but that doeant mean TB4 can be open for all.

The security issues are common to modern CPUs. The Spectre/Meltdown type vulnerabilities relate to speculative execution and I think that some of the recent issues relate to hyperthreading. One way to kill the hyperthreading involves turning it off. But nobody wants to give up the performance benefits of hyperthreading.

 

[cube]

Intel CPU sucks and having tons of security issue. I dont mind if Apple makes ARM MacBook series. The only concern would be apps like Adobe.

Apple could just switch to Zen.