Initial Apple M1 SoC Support Aims For Linux 5.13 Kernel

[lanjoe9]

Why should someone do this work?
Is it so much fun?

As surprising as it is, some people really do like this sort of work (reverse engineering, testing, finding hardware secrets and implementation quirks).
Not to mention that Marcan42 is getting paid for that.

[smitty3268]

From a silicon point of view first 8086 was released in 1978 while the first ARM was released in 1985. In both cases design started about 2 years before.

If we talk about later variants of the ISA, AArch64 vs x86-64 then we are talking 2011 vs 1999.

EDIT: oops AdrianBc and morydris already pointed this out

ARM is certainly newer, but I don't think it's particularly accurate to just call x86 legacy and old, vs ARM as modern and current. ARM has been around for a while too.

Plus "x86" cpus aren't really even based around x86. They have their own internal ISA's that are RISC based and just have an extra bit of silicon to translate the x86 code into their internal ISA, which can vary from generation to generation as much as AMD or Intel want it to. It is an extra bit of cruft these CPUs have to carry around, but it's not like these are CPUs with a 1980's design and it's misleading to portray them that way.

[ldesnogu]

ARM is certainly newer, but I don't think it's particularly accurate to just call x86 legacy and old, vs ARM as modern and current. ARM has been around for a while too.

I was only giving some data points about releases of ISA and CPU. AArch64 is much more recent than x86-64. And a lot of older features of ARM have been removed (like having the flags in the PC). As far as I know x86 chips can still run in 16-bit mode. AArch64 is definitely much cleaner than x86-64.

Plus "x86" cpus aren't really even based around x86. They have their own internal ISA's that are RISC based and just have an extra bit of silicon to translate the x86 code into their internal ISA, which can vary from generation to generation as much as AMD or Intel want it to. It is an extra bit of cruft these CPUs have to carry around, but it's not like these are CPUs with a 1980's design and it's misleading to portray them that way.

ISA and implementations of an ISA are two different things. Anyway no one cares about the "internal ISA", no one can program it or generate code for it. It's only here to be able to run fast the abomination x86(-64) is.

[sarmad]

Nice. So hopefully by 2030 we'll have usable support for the M1.

[smitty3268]

I was only giving some data points about releases of ISA and CPU. AArch64 is much more recent than x86-64. And a lot of older features of ARM have been removed (like having the flags in the PC). As far as I know x86 chips can still run in 16-bit mode. AArch64 is definitely much cleaner than x86-64.

I was referring back to the original post that started this thread.

ISA and implementations of an ISA are two different things. Anyway no one cares about the "internal ISA", no one can program it or generate code for it. It's only here to be able to run fast the abomination x86(-64) is.

No one really cares about external ISA either, though. The ISA is pretty meaningless for 99% of developers who will just rely on the compiler to handle all that for them. All they care about is the performance of the chip.

[WorBlux]

And not just that, the x86 ISA is not even handled directly anymore, its commands get split up in smaller tasks and executed by RISC type cores in all x86 CPUs since the Pentium Pro.

1. The extent of micro-ops in x86 is often exaggerated. And Intel's gracemont doesn't do it at all
2. Newer ARM designs feature both micro-ops and instruction fusion.

It's got as much to do with how you lay out your execution pipelines than the ISA.

[Slartifartblast]

I can't see Apple wanting you to step outside their walled garden, I wonder if they'll turn the screw further to make it either harder or even impossible to run anything else on their systems in the future.

[Boland]

I can't see Apple wanting you to step outside their walled garden, I wonder if they'll turn the screw further to make it either harder or even impossible to run anything else on their systems in the future.

They just made it EASIER to run alternative operating systems with the latest OS update. If they didn’t want people to run Linux they wouldn’t have put in features that let that happen.

[WorBlux]

I can't see Apple wanting you to step outside their walled garden, I wonder if they'll turn the screw further to make it either harder or even impossible to run anything else on their systems in the future.

If nothing else, it's an anti-trust defense. To leave other-OS and non-app-store installs possible, though difficult+non-obvious gives them something to point to as competition.

Big Sur doesn't lock out an "anywhere" install, but does hide the feature by default. Both enterprise and the Unix crowd are too entrenched in the OS for it to go to a complete lock. And they've made enough public statements about it that behavior to the contrary would be legally problematic in other ways as well.

[wizard69]

I was referring back to the original post that started this thread.

No one really cares about external ISA either, though. The ISA is pretty meaningless for 99% of developers who will just rely on the compiler to handle all that for them. All they care about is the performance of the chip.

Actually in many cases the processor doesn't even matter as again the developers just rely upon the compiler and the OS to take care of everything. I was rather surprised how quickly Homebrew got the key open source software running native on M1 Mac OS. A lot of that was already ported to ARM Linux which probably helped, the only real surprise how well that software ran so early in the process.

What makes M1 and ARM in general so appealing is that you can get " as good or better" performance at a fraction of the power used. Sometimes it makes me wish that Redhat would get into the hardware business. I could see significant interest in a well supported ARM based laptop that is as good as what Apple is shipping.