Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
ARM based MacBook Pro discussion [merged]
- Thread starter Omega Mac
- Start date
-
- Tags
- arm intel macbook pro macos sierra
- Sort by reaction score
The new version is Windows 10 on ARM which is similarly languishing from an Intel/X86 replacement perspective.
AFAIK the only issues are Intel's current licensing and certification policy rather than anything technical... although many current ARM products don't use PCIe (Thunderbolt is basically external PCIe) in which case its kinda moot.
"USB4" sounds like the resolution of that, and although its not 100% clear with the details currently available but it sounds rather like USB4 is going to include a 'generic' version of TB3, but only manufacturers signed up with Intel will be able to use the "Thunderbolt" brand.
Long term, Thunderbolt probably needs ARM and AMD more than ARM and AMD need Thunderbolt. TB3 is still "niche" compared to USB so I doubt Intel could use it as a hostage to keep the likes of Apple stuck on Intel CPUs - Apple already have the iPad Pro and if they can't add TB3 to that in the future they'll find/design an alternative interface.
I'm sure Intel will make sure they have a healthy licensing income stream from "USB4" - but we're talking about the likes of Apple and ARM here, not some Mom & Pop open-source project - paying each other to license technology is standard operating practice, we muggles only get to hear about it when it goes all Quallcom.
1984: If devs have to re-write their Apple II software for Mac, , they'll just stop developing
1994: If devs have to re-compile their 68k software for PPC, they'll just stop developing for Mac.
2001: If devs have to re-write their software for OS X, they'll just stop developing for Mac.
2005: If devs have to re-compile their PPC software for Intel, they'll just stop developing for Mac.
2018: If devs have to re-compile their 32-bit software for 64-bit, they'll just stop developing for Mac.
...and all of the time the proportion of software written in high-level languages and using hardware-independent OS frameworks for graphics/sound/acceleration - for which 're-compile' really does just mean "tick the 'ARM' box in XCode and hit 'build'". Meanwhile, anything not being actively maintained is going to die with the next version of MacOS anyway when it drops 32-bit support.
Linux - along with the vast majority of the well-known Linux/Un*x packages - run quite happily on ARM64. You'll need a hypervisor of course and Parallels/VMWare are two things that might not make it to an ARM Macintosh - but there's a hypervisor framework in MacOS which, presumably, Apple will port. I doubt that you'll have to wait long for something like Docker (it runs on ARM, the Mac version uses the MacOS hypervisor framework) to appear.
OK, so I know that 'runs on ARM' doesn't mean 'MacOS for ARM version available on day 1' - but it does mean "doesn't depend on a huge blob of lovingly-handcrafted x86 SIMD assembly code written by a guy now living on top of a pole in Peru". Apple could kill the Mac by trying to force the change-over too quickly, or by using a switch to ARM as an excuse to lock-down the Mac iOS-style (but they can do that anyway on Intel as soon as every current model has a T2 chip).
...but assuming they do it on a sensible timescale and don't lock it down, I think the only long-term casualty will be the ability to run x86 MS Windows which will be a problem for some - yes, there's Windows 10 for ARM but there's no indication that MS will make it available for purchase and I kinda suspect that people who need Windows do so because they have x86 software to run. Personally, though, while Windows VM was a killer app 10 years ago, I'm finding that less and less important (particularly with the near-death of Internet Explorer).
A switch to ARM wouldn’t impact my life very much. In the event that I wanted to switch to an ARM-based Mac, my app requirements are pretty low.
Terminal/SSH access, a web browser, standard email client, and some sort of Word and Excel compatible solution will cover about 90% of my needs. For the other 20% (I always give 110% ... always,) I can use something like Splashtop to remotely access a VM at home if needed.
Terminal/SSH access, a web browser, standard email client, and some sort of Word and Excel compatible solution will cover about 90% of my needs. For the other 20% (I always give 110% ... always,) I can use something like Splashtop to remotely access a VM at home if needed.
If Arm chip with desktop os were viable and vastly better than intel, we shouldve seen so many product running win10 machines already. As I stated in other thread, the performance of Arm chip must be tested with real desktop usage with real people.
After it is proven, then I will believe.
After it is proven, then I will believe.
The TB4 spec is merging with USB4 so this *shouldn't* be a problem. Apple will just jump to USB 4.0. The other advantage custom Apple silicon could bring is more PCIe lanes. Intel is starting to improve but even the 2013 Mac Pro was hampered by not having enough PCIe lanes to properly implement TB2! One reason that machine languished for so long is that Intels updates to that Xeon line didn't include any more PCIe lanes and so an update to TB3 wasn't even possible.
"The TB4 spec is merging with USB4 so this *shouldn't* be a problem"
Wrong. TB4 is not even announced yet and why TB4 should have the same speed as TB3? Also, how can you guarantee that Intel will allow Apple to use TB4 in the future?
Great but that isn’t my point.
If I can’t mirror the environments in production it’s useless.
I’ve had projects where it’s an issue because someone has a cheap-ass windows laptop with a 32bit CPU and the packages we used weren’t necessarily the same for 32bit vs 64bit.
...and is that still going to be an issue in ~5 years' time when support for Intel Macs might actually dry up? Even now you're using the past tense. Will VMWare/Parallels/VirtualBox still support that Linux distro in 3 years time (probably after support for the distro itself has ended)? Meanwhile, even an Intel Mac can't emulate the performance, quirks and specific hardware of your target environment.
I can't actually speak for your case - it may be that you will need to run particular x86 virtual environments for the foreseeable future, and a switch to ARM Macs might send you off to PC land... but then, one of the reasons that Mac is nicer than Windows is that it hasn't been hobbled by the "OMG if we drop support for 16-bit code then United Holdings Inc. won't be able to run their 30-year-old payroll software and we'll lose a 10,000 user annual support contract!" factor.
However, I think people are projecting a 2005 problem on to 2020. In 2005 if you were producing, say, a web app, then you'd have to religiously check every last step across Safari, Firefox and a couple of versions of Internet Explorer - with the latter needing a shedload of browser-specific work-arounds - or risk going down a blind alley that simply would not work on some browsers. 2019, and although the need for cross-platform testing hasn't gone away entirely, it has greatly diminished and the issues are a lot more subtle. The last couple of issues that came up on VM testing turned out to be red herrings caused by the hypervisor, and the more pressing issue is checking that UIs work on iPad and Surface touch screens - its getting to the point where a cheap Windows laptop or 2-in-one would be more useful than a VM for testing.
(The 2005 vs 2020 argument goes for 'porting' stuff to ARM, too - modern code is less likely to be architecture-specific - it was the WIN16-era stuff that was really horrendous... remember "near" and "far" pointers? Unix/Linux has been built on cross-platform code wherever possible since t=0).
I'm currently writing some code that runs on both Mac and Raspberry Pi - the actual coding differences are trivial (and to do with Darwin/BSD vs Linux, not architecture) - the things that need testing are to do with the physical hardware (the Pi processor is pretty fast but it has a horrible I/O bottleneck problem).
Server-side - Its 2019 and if you need a temporary Linux server for testing just log into (say) Linode and pay $5/month (billed by the hour!) to spin up a container (or Amazon AWS/whatever) - we have WiFi and by the time that nice Mr Musk has his newly launched satellites up and running you'll have broadband on the top of Everest. The majority of server-side development is using hardware-independent scripting languages anyway and all of the usual suspects (apache, ngenix, node, mongo, docker, mysql, postgresql, python, Java, PHP...) are already up and running on ARM linux.
For serious native Windows/Linux application development - frankly - if I was doing a lot of it I'd just get a PC. It's not like they're expensive.
I'm not saying that VM is dead - its still a useful tool - just that its maybe not the killer app it was in 2005 when web development and platform-independent tools were less prevalent. Even for users, a lot of the PC-only tools and websites that required Internet Explorer are now dead and replaced by Apps or standards-compliant websites.
Heck, even if Apple announces ARM next month, by the time Intel macs actually become obsolete I might even have finished moaning about USB-C
[doublepost=1558703280][/doublepost]
OTOH - as you say - TB4 is not even announced yet, so why are people worried about "what if Intel brings out TB4"?
Although its not 100% clear, it does sound like "USB 4" is likely to include a generic, plug-compatible version of TB3.
Apple have been a key promoter of Thunderbolt in the past - in the PC world its something of a solution looking for a problem*. If Apple don't/can't support a hypothetical TB4 because Intel try to make it exclusive to their CPUs then it's probably going to hurt Intel more than Apple.
Remember, even if Apple don't move the Macs to ARM they're already pushing the iPad Pro as a serious laptop alternative - the point will come where, if they can't put the latest Thunderbolt on the iPad, Tim Cook will be standing up at WWDC 2020 and explaining why you don't need Thunderbolt.
* especially as a video/eGPU interface - the PC world has this silly thing about (a) having video outputs that output video without moshing it together with a lot of unconnected protocols and (b) putting powerful GPUs inside the computer with a 16-lane PCIe connection rather than in an external box that adds $300 to the cost of the GPU and only has a 2-lane interface. Mad.
You’re missing the point.
The whole point of running a local virtualised environment on an amd64/x64/whatever nomenclature you prefer is that it’s identical to production. Being 32bit rather than 64bit was one example where that’s an issue, being arm vs amd64/x64/etc would be too much difference to realistically expect reliable results.
I don’t know what you work on or how, but the amount of time I spend writing code without a VM running in the background is negligible. A VM on the other side of the planet isn’t really a replacement for that.
A Windows or Linux “desktop” would just mean that I’m running VMs on Windows or Linux, and having supported other developers who use both those setups: forget it.
Like I said, if they drop support I’ll likely end up with a Xen or ESXi host in the office, but it’s a step back in portable functionality.
And, this whole premise of “just use a vm on arm” is predicated on the arm cpus having hardware support to make virtualisation performant.
Well said!
The whole problem here is people very much forget that technology isn't like most other industries. You can't just swap part A for part B and it all works the same way and just because one component works great over here doesn't mean it will over there.
If anything, computing on "desktops" (and that includes laptops as well) is becoming much more specialized since regular users are transitioning more and more toward the tablet and smartphone markets. As a result, it isn't enough to be just able to run Chrome...
I'm a bit sad to see so many people totally gung-ho with the Mac moving to a new architecture that is incompatible with the rest of the PC industry.
One of the Mac's biggest strengths right now is that compatibility. The Mac is, basically, an x86 EFI PC. It can run other operating systems that run on regular PCs. Heck, regular PCs can run MacOS with some finagling, though Apple isn't too big on encouraging that.
Do we REALLY want the Mac to turn into a proprietary, closed architecture system like iOS devices? If you want that, you can buy an iPad or iPhone and enjoy an Apple ARM system to your heart's content.
Let Macs continue to be industry-standard EFI PCs. They are a lot more flexible and capable as a result. We will miss that A LOT if Apple moves to ARM on Macs.
One of the Mac's biggest strengths right now is that compatibility. The Mac is, basically, an x86 EFI PC. It can run other operating systems that run on regular PCs. Heck, regular PCs can run MacOS with some finagling, though Apple isn't too big on encouraging that.
Do we REALLY want the Mac to turn into a proprietary, closed architecture system like iOS devices? If you want that, you can buy an iPad or iPhone and enjoy an Apple ARM system to your heart's content.
Let Macs continue to be industry-standard EFI PCs. They are a lot more flexible and capable as a result. We will miss that A LOT if Apple moves to ARM on Macs.
Well said!
Quite honestly if the Mac platform moves to arm I’m not really sure why I need to use it anymore.
That would probably put me into a future of real heavy lifting work being done on a normal PC still and I would probably just also still have an iPad and iPhone on the side
There are bunch of Debbie Downers on here. I'm not worried at all. A lot of people moved to that post-PC device — the iPad or a Chromebook. I'm seeing them in lots of places. So as far as I'm concerned let Apple transition Macs over to ARM processors. It seems more than inevitable at this point. With the iMac and MacBooks at the precipice of redesigns, a new display and the Mac Pro coming shortly, along with the integration of UIKit in the Mac that opens the Mac up to the plethora of iOS apps, the future looks really bright for the Mac for the first time in a long time. Let's go!
Just because you see them everywhere doesn’t mean they’ve replaced PCs! ARM just isn’t powerful enough. If they were better than x86, we would have moved to them a long time ago.
ARM is for power efficiency, AM64/EM64T is for performance
I generalized and wrote TB4 instead of TB3. These are very recent announcements. TB3 and USB4 will converge. Here's one new story: https://www.extremetech.com/computing/286861-thunderbolt-3-rebranded-as-usb4-coming-2021
A little search and you'll find plenty more.
Lol, I know that story for a long time. Why do you even mention the unrelated topic in here for? The point is AMD CPU is still not able to use TB3. USB 4 will take time and TB3 will have more supports than USB4 because Intel created TB technology.
What if Intel made TB4? How Apple and AMD suppose to get TB4 without Intel CPU? This is the point that I've been saying. Read before you comment.
Well, Intel created USB as well...
USB4 will be widely supported including on AMD devices as USB4 is Intel finally living up to its promise to open TB up to the industry. That doesn't mean every part of USB4/TB3 will be required or even supported everywhere. For instance, ARM implementations won't support PCIe I am sure...
Intel opened TB3 ONLY. Did they ever said the future TB will be available to all brands? NO. The point is the future TB may not be available to AMD and ARM just because of Intel. USB 4 will be available after 2020 or maybe in 2021.
ARM is totally all in one computer at this point. I cant even expect PCIe anyway because it prioritizes for laptop line ups
I did read, did you? Seriously go back and read the link. With USB4, TB3 will essentially be rolled into the new standard. Yes, I misspoke and said TB4 but chill and learn some manners.
[doublepost=1559266682][/doublepost]Also, AFAIK, the licensing agreement between Apple and Intel over Thunderbolt isn't public. Yes, Intel developed a lot of the foundational tech but Apple stepped in and influenced the final implementation (remember the optical USB-A port?). We don't know what the nature of that relationship/agreement is and Apple is free to modify many aspects of the ARM architecture to add IO that suits their needs (PCIe, modems, etc.).
While I don't know for certain that it ISN'T an issue I'm not convinced that it would/will/is keeping them from switching to ARM.
Not me. One of the best things to happen was Apple switching to intel. Being able to run VMware and various x86 OSes has made me more productive.
In the PPC days I tried VirtualPC but it wasn't great.
I'd have to wait and see how things turn out if they do switch to their own ARM chips, but it is very likely that my Apple devices will be limited to my phone or tablet.
In the PPC days I tried VirtualPC but it wasn't great.
I'd have to wait and see how things turn out if they do switch to their own ARM chips, but it is very likely that my Apple devices will be limited to my phone or tablet.
You do know that Intel makes multiple CPUs of varying power consumption and ranges? I assume English isn't your first language and that is why you don't pick up on the nuance.
Nobody forced Apple to use high end chips that generate a lot of heat in a tiny package. They could have either designed a better chassis and cooling system OR limited themselves to the lower power chips (like the Macbook).
Apple made the decision to put those in a tiny package, not Intel.
Take your own advice and Google it. You can learn about the range of chips available from Intel.