Running an alternative OS "natively" on a Mac doesn't seem very appealing to me. It's almost guaranteed you will never fully get Wi-Fi, audio, graphics acceleration, the trackpad, etc. working flawlessly, especially on the M1 where almost every component is Apple proprietary and it's not a matter of loading, say, some common Intel graphics driver.
On the other hand, virtualization _already_ lets you run whatever arm64 BSD, Linux, or Windows that you want. The trackpad and everything else will work flawlessly thanks to guest additions. Yes, you make some tradeoffs -- you sacrifice a core, a few GB of RAM, some disk space for an OS you don't _really_ want to be running. But I'm skeptical that these are not better tradeoffs than constantly using a hacked-together and half-broken OS.
To me the ideal solution is not that Corellium or Marcan or whoever reverse engineers the drivers to a sufficient degree. I'd rather see Apple (or the Hackintosh types) release a tiny variant of macOS that contains hardware drivers and Hypervisor.framework and not much else. It could be the next generation of Boot Camp and it would probably take less engineering effort for them to maintain while satisfying the small segment of customers interested in running other operating systems.
> On the other hand, virtualization _already_ lets you run whatever arm64 BSD, Linux, or Windows that you want. The trackpad and everything else will work flawlessly thanks to guest additions. Yes, you make some tradeoffs -- you sacrifice a core, a few GB of RAM, some disk space for an OS you don't _really_ want to be running. But I'm skeptical that these are not better tradeoffs than constantly using a hacked-together and half-broken OS.
True in theory, but at least with Linux I've found that desktop environments exhibit a slight but maddening level of sluggishness in VMs, no matter if it's GNOME, KDE, XFCE, or MATE. If all you need is a commandline Linux VMs are great but I'd go insane trying to use one as a graphical desktop for an extended period.
Really, that goes for any guest OS other than maybe Windows, and Windows is different only because VM developers have poured an enormous amount of time and money into it.
You can get rid of the sluggishness with GPU passthrough... but then the guest OS would still need GPU drivers.
I'm don't think you can do GPU or PCI passthrough on macOS. Not sure if/when Apple would consider adding that.
I'm glad the things I want to do on Linux are all terminal based. I used a tiling WM for years on Arch and don't miss it much with all the macOS WM extensions available now (Magnet, Rectangle, Moom, Hammerspoon, etc). Less themeable but perfectly functional.
I doubt Apple has any intention of publishing hardware documentation for someone to write a native driver, but there are behavioral reverse engineering efforts already underway.
Very nice. An advantage of paravirtualization is that compositing across the guest and host will work, unlike passthrough. Any rough indication of the level of performance?
It would be to Apple's advantage somewhat to make that happen right? Apple wants to sell hardware and does not mind people running virtualised Windows or Linux on it?
They also sell services though, and their products reinforce one another so that once you’re bought into their ecosystem it ties you to Apple products in perpetuity. I know because it happened to me; doesn’t happen if you’re not actually using the OS.
>so that once you’re bought into their ecosystem it ties you to Apple products in perpetuity.
Only because of the seamlessness (within reason) of the experience.
Otherwise, there's nothing to switching to another OS for all or any part of it.
Many (most?) apps people use everyday are web- or Electron based, so they can switch to any other Windows or Linux with e.g. Chrome.
Their music is in some streaming service (Spotify, Apple Music, Google Play, etc) all of which play on other OSes (and in any case, they can change subscription and they haven't lost anything). Or, for the fussier, it's mp3 or flac files, they can easily move. Even bought music (aac) from the iTunes Store is playable in other platform (since Apple has removed the DRM for a decade or so).
Ditto for movies. It's all streaming nowadays, and you can just switch to another streaming service (and all play everywhere anyway).
As for professional apps, most heavy ones (MS Office, Adobe Suite, Cubase, Live, Resolve, etc) run just fine in both OSX and Windows, and come with dual installers, so you can switch if you like (some even allow for multiple computers at the same time).
(If you're using Logic or FCPX, then sure, you'll need to find another program, and go through the pain of migrating your projects).
For mobile it's the same thing. E.g. you can sync just fine with an Android phone (you just don't get niceties like Handoff, Scan-to-Notes, etc).
So there's nothing special to "tie you to Apple products in perpetuity" except "I might lose the use of some apps that are Mac Only" (so same thing that would have tied you to Windows or Linux in perpetuity).
Apple also tends to prefer pushing their users to their closed software ecosystem. Given the likely low demand for passthrough on these devices, I kind of doubt they'll implement it in the foreseeable future.
There are a couple of projects that improve this (for linux host + linux guest). virgl exposes accelerated graphics to the guest, which chromeOS uses in its offering. Intel also has its GVT-g offering which virtualises intel integrated GPUs. Of course, none of this applies to apple.
> Running an alternative OS "natively" on a Mac doesn't seem very appealing to me. It's almost guaranteed you will never fully get Wi-Fi, audio, graphics acceleration, the trackpad, etc. working flawlessly, especially on the M1 where almost every component is Apple proprietary and it's not a matter of loading, say, the right Intel graphics driver.
It's not appealing to me now. In 9 or 10 years when Apple has stopped shipping new versions of MacOS for the platform, it will be much more interesting for me. Of course by then it's just a curiosity or fun project, not something I need for day to day. And by then, much of the wifi/ graphics/ audio issues will be solved.
I wouldn't be so sure that wifi/graphics/audio will ever be solved. It's possible, but not guaranteed. There are whole generations of Intel Mac systems that have never received support for these.
This statement is a bit presumptuous: "It's almost guaranteed you will never fully get Wi-Fi, audio, graphics acceleration, the trackpad, etc. working flawlessly."
Maybe I do not want all those things every time I boot. Apple used an "alternative OS" for their Airport routers: NetBSD. Kernel was around 6.x if I recall correctly. Seemed like Wi-Fi worked preety good, maybe even "flawlessly".
It's a computer. As the purchaser, I should get to decide how it is used. Using Apple software should be optional.
"Presumptuous" meaning it presumes what a user will want to do with a computer. In this case, the computer comes with no display, no keyboard, no mouse, no trackpad, etc.
I'm not following. Are you saying that one can use an Apple router with an alternate free operating system and it still functions as a router ? I'm not aware of any such project. At least openwrt does not support the Apple router. As for what users would want to do, that is a moot point since there are common expectations for devices based on the specs and form factors. So, why would you buy a $2k laptop to use it as an inefficient, buggy, and underpowered headless linux server (assuming you get networking working)?
The "alternative OS" is the OS that Apple used for its AirPort Extreme router, and which it borrows heavily from for the MacOs userland. Apple writes some software itself, but it also takes a significant amount of source code straight from "alternative OS" projects. To be clear, the OS projects I am referring to are not Linux nor is the computer that is the subject of this thread a "$2K laptop".
Do you mean to imply that those things are known not to work?
There have been many Apple devices where all those mentioned things (if present) work under a non-Apple OS. So as usual the truth is somewhere in the middle. From my Linux experience there are a lot of well working MBP models from this time span. And Minis.
I'd like to point out that the dmesg indicates a M1 Mac Mini. There is no display, no keyboard, no mouse, no trackpad, etc. I think there may be a 3.5mm headphone jack, though.
I'm just really curious and don't have a dog in this fight against/for ability to do as you will with a product you purchase.
A) a hypothetical device, let's not call it a computer, but similar, exists that asks you politely by means of something like a license not to modify it even though it would be possible to do so pretty easily.
B) another device, this time a hypothetical CPU that is next to impossible to modify with any degree of ease since it requires nanoscale tools.
Why should there be a fundamental difference in a person's ability to modify one of the above, but not the other? Just because the laws of Physics don't allow mere mortals to so (not at least without a highly sophisticated nano fabrication lab) doesn't seem like a very convincing principled answer. Is there a stronger argument?
Intent is important in law and in most people's moral principles (acknowledged or not). It seems to me that going out of your way to make the things you sell easier to repair is supererogatory, but going out of your way to make the things you sell harder to repair is bad and wrong, even if two companies did those two different things and reached the same end result.
>Why should there be a fundamental difference in a person's ability to modify one of the above, but not the other?
Because one could trivially modify the first, and people do all the time (and screw the law), whereas the second is, as you said, "next to impossible" to modify even if you want it.
>Just because the laws of Physics don't allow mere mortals to so (not at least without a highly sophisticated nano fabrication lab) doesn't seem like a very convincing principled answer
In what universe do "the laws of physics prevent this to be done easily without millions of dollars" is not a convincing answer?
> Apple used an "alternative OS" for their Airport routers: NetBSD. Kernel was around 6.x if I recall correctly. Seemed like Wi-Fi worked preety good, maybe even "flawlessly".
That’s a nice strawman you got here. AirPort routers shipped with NetBSD, and it was officially supported. But good luck getting macOS to run on that — macOS would count as the “alternative OS” for the AirPort.
Being able to run OpenBSD or FreeBSD on a M1 mini would be AWESOME. That little machine is pretty powerful and does not need much power to do what it does.
If you are talking purely about desktop, I have to agree with you that the only real option is MacOS because of all the apple-only stuff (at least for now until drivers show up)
Most laptops are more or less in the same situation as M1. Maybe they different quantitatively, but not qualitatively: devices do not have (sufficient) specs available and people need to reverse engineer them if you want to run any free OS on them. Yet people want to do that and often manage. Why is M1 different?
What I really would like to see is hardware manufacturers (Apple and anybody else) releasing the specs, so that people are free to do whatever they want with the metal they buy. I might event consider appropriate a legislative intervention in this field.
Not all laptops. Regular stock Intel components, for example, have solid Linux support because Intel nowadays hire kernel developers to support them. Some even have upstream drivers before the hardware is even commercially available.
Anecdotal counterpoint: I bought a cheap Intel SoC laptop in 2019 thinking it would make a nice ultraportable writing machine. Things that it shipped without Linux support for: the trackpad, wifi, Bluetooth, and disk (eMMC controller). It just about works now with the latest kernel version and some manually compiled drivers, nearly two years later.
It doesn't matter much for laptops which increasingly seem to depend on Intel/AMD integrated graphics, which are more capable than ever in laptops with more stringent size and power constraints. How many MacBooks sold have discrete graphics cards by any manufacturer? I'd like to see data on this but I couldn't find any relevant statistics after a bit of searching.
>>Running an alternative OS "natively" on a Mac doesn't seem very appealing to me. It's almost guaranteed you will never fully get (...) working flawlessly
This statement is rather miopic. I still recall when this very same argument was made regarding the idea of wanting to install Linux in computers which were bundled with Windows. I can tell you it made absolutely no difference to me a couple of decades ago when I installed Mandrake Linux on a Windows laptop, and it made no difference to me that I couldn't get hardware acceleration or the webcam to work, and that wifi only worked with ndiswrapper. I was able to run the software I wanted to use in the hardware I bought, and that's what matters.
M1 Mac minis are on the market for around 800$, which is an affordable price point, and their form factor and looks make them an interesting choice for a home server or workstation. It would be incomprehensible if we couldn't take advantage of that hardware by running our choice of software.
The iMac G4 didn't support virtualization, so booting an alternative OS is more attractive. And the Nvidia GPU is probably easier to find cross-platform drivers for. New drivers or some sort of IOKit compatibility layer need to be developed for the M1 GPU (which is orders of magnitude more complex).
(I can run OpenBSD on any modern piece of hardware, though; the novelty and nostalgia are in running Mac OS 9 on that.)
Yeah it's a great model for classic Mac gaming as it has pretty good specs, 3d accel, and can run OS 9 and OS X. I've got two of them (among many other Macs) so I figured I'd try OpenBSD on this one, haha :)
I run Linux natively on a MacBook just fine. In fact I had to, when High Sierra went out of support not that long after I’d upgraded my RAM and HDD to SDD. It now flies and honestly I don’t know why I waited until High Sierra went end of life as I feel Mac OS X has got worse and worse. Also the hardware has got less and less upgradable/fixable in that time too.
> On the other hand, virtualization _already_ lets you run whatever arm64 BSD, Linux, or Windows that you want. The trackpad and everything else will work flawlessly thanks to guest additions.
The OpenBSD kernel doesn't support loadable kernel modules, so virtual machine guest additions are not an option for a virtualized OpenBSD.
>It's almost guaranteed you will never fully get Wi-Fi, audio, graphics acceleration, the trackpad, etc. working flawlessly
And yet people have done it with Linux just fine... Linus himself run a G5 as his desktop (circa 2003-2005) and later a Macbook Air as his daily driver laptop (circa 2011-2013...)
I think there is another way to look at it. If not for Apple’s hardware policies, you would just be able to run your favorite OS on their hardware. It’s the fact that they publish EULAs and NDAs so long that it probably wouldn’t have fit I to the RAM of their first computers which actively slow down or prohibit any work by developers. Why isn’t it Apple whom we should find unappealing?
On the other hand, virtualization _already_ lets you run whatever arm64 BSD, Linux, or Windows that you want. The trackpad and everything else will work flawlessly thanks to guest additions. Yes, you make some tradeoffs -- you sacrifice a core, a few GB of RAM, some disk space for an OS you don't _really_ want to be running. But I'm skeptical that these are not better tradeoffs than constantly using a hacked-together and half-broken OS.
To me the ideal solution is not that Corellium or Marcan or whoever reverse engineers the drivers to a sufficient degree. I'd rather see Apple (or the Hackintosh types) release a tiny variant of macOS that contains hardware drivers and Hypervisor.framework and not much else. It could be the next generation of Boot Camp and it would probably take less engineering effort for them to maintain while satisfying the small segment of customers interested in running other operating systems.