Mac Intel -> ARM switch

As a language hacker myself, I’m curious as to people’s thoughts (Xojo engineer’s really I suppose) as to what would happen to Xojo on the Mac if Apple decided to switch from Intel CPUs to an ARM-based architecture. There are plenty of rumours that 2019 could be the year that Apple make the switch so it could come sooner than we think.

Is this something that Xojo could handle swiftly or are we talking a massive undertaking? I’m guessing that using LLVM at least makes the switch feasible for Xojo?

Anyone have any thoughts about this? I know it’s only speculation but sometimes that’s fun!

I think they are largely ready for it, the Raspberry Pi is ARM-based and the Ipad’s too. The biggest problem probably will be the transition to 64 bit.

Xojo for iOS is 64 bit only, as is iOS itself.

They’ve already transitioned to 64-bit on all platforms with the LLVM compiler so I’m hoping it’ll be straight forwards.

Personally I’m hoping Apple decide to drop Intel for AMD in the short term (especially since I’m waiting for them to release a new iMax). The Ryzen Threadripper chips are pretty special.

One of the reasons we went with LLVM was to make future transitions like this easier on us. But I am not commenting on what Xojo might do in a possible response to Apple ARM Mac rumors, as fun as those rumors might be.

Slightly off topic, but I sincerely hope they do not do this. For a decade now MacBook Pros have been ideal developer machines. Mac, Windows, Linux, UNIX, mobile Android, and mobile iOS (granted the last two are emulated; but the dev tools are x86). Just launch a VM for the other OSes and you’re good to go. Any office I visit, any meeting with any client, any project I’m doing…everything is on my MBP.

I’ve run into developers who are not primarily interested in macOS or iOS yet use MBPs for their work.

A switch to ARM would kill the ability to host multiple OSes on one notebook. The ideal of having all your projects, files, and test environments on one machine would be dead. I hate the thought of having to lug around multiple machines. And quite frankly macOS is the best VM host for dev purposes. I don’t want my primary/base OS to be Windows or Linux, nor do I want to try and VM a semi-legal copy of macOS on a Windows or Linux notebook.

I remember the RISC v. CISC wars of the 90s. Backwards and cross compatibility has trumped every RISC advantage. The only reason ARM succeeded in mobile is because there was nothing to be backwards or cross compatible with, so efficiency won. But x86 has literally 41 years of legacy desktop and server code. There’s zero reason to walk away from this.

Now if they want to start exploring AMD processors on the desktop…I’m fine with that :slight_smile:

What can be this for you, not, I suppose.

I dread the thought of Apple Dropping Intel. What happens to all the Intel Optimized Code . SIMD, MMX etc…
Bad enough nVidia is not in every mac because there goes CUDA.
Apple is a little bit nutso if you ask me.

Well, they could simply support two CPUs and see which is better over time.
ARM for small mobile MacBooks.
And Intel for the workstations yet to come.

On the other side, they could ship both and have two processors working together.
As of now the T2 in the MacBook Pro already does some work like video encoding.

Didn’t already support two CPUs (iPhone / Macs) ?

I think they already know.

Once the price will be involved, the decision will be taken, no matter what technics will be / how hard that could be.

They already do that once (or twic / thrice ?).

The jump from 68K to PowerPC was made for long term performance reasons. The jump from PowerPC to x86 was made because personal computer marketshare was so small for PowerPC that Apple couldn’t get timely updates from IBM or Motorola. Motorola was more interested in embedded and IBM was more interested in workstation/server. Apple could get big, hot, power hungry chips from IBM for desktop (i.e. G5) but nothing for mobile. When IBM failed to even deliver a planned clock speed jump for the G5 it was over.

This. ARM may be faster/more efficient at some price points, but we will lose all the optimized x86 code on top of losing the incredible level of compatibility we have today. It would be a very foolish move.

Ironic iPhone autocorrect…

I would be astonished if Apple don’t make the jump to ARM within the next 12-24 months. At least in their low end notebooks. The new iPad Pro already outperforms the 13” MacBook Pro in GeekBench.

If they insist on staying with x86 though I am curious as to why they have held off of AMD’s CPU offerings. Their new Ryzen architecture is leaps and bounds better than their previous offerings and they can really jam a high number of cores on the die. Of course, multiple cores doesn’t really help the Xojo IDE.

When I woke up this morning, I had a revelation: you meant iMac !

Back to the Intel ARM switch: the new ARM CPUs will outperform Intel CPUs for some years, and I cannot believe that a move is reasonable until a brand new architecture (way to build a CPU) saw the lights.

Only the future will tell us what will happens.

By 2%. Power Macs were typically 50-100% faster than their x86 counterparts in the 1990s. x86 still won. Back and cross compatibility trump performance.

Agreed. I wouldn’t mind seeing AMD chips offered in the Mac lineup.

Wouldn’t it help 64-bit compile times?

Xojo doesn’t leverage multiple cores.

It absolutely does when you compile as 64-bit or ARM.

Oh that’s cool, is that because of LLVM?
Built apps like the IDE certainly don’t use multiple cores. So I suppose I was wrong by quoting the compile times note.

Left Windows because they wanted to corrupt java and the web.
Are we to leave Mac because they wont standardise to a public API?
Looks like a CUDA-Gras for nVidia.

Only the compiler uses multiple cores right? To speed up compilation time? I was under the impression that built apps only ever run on a single core. Very happy to be corrected.

AFAIK it’s just the compiling process that uses more than one core, not the created application.