New Patreon project seeks to bring Linux to M1 Macs

Posted:
in Current Mac Hardware edited November 2020
Developer Hector Martin, also known as "marcan," on Monday launched a Patreon to fund solo work on a Linux port for Apple silicon Macs.

MacBook Pro


According to Martin, Apple's M1-powered Macs are capable of running Linux, but creating a working port is a major undertaking; a near Herculean effort for one developer. He says he is up to the task, though the project would require dedication equivalent to a full-time job, thus the Patreon.

Martin has experience with open source code, from projects involving Nintendo's Wii console to more recent Linux ports for Sony's PlayStation 3 and PlayStation 4. The Patreon's goal, according to the developer, is to massage Linux on M1 Macs "to the point where it is not merely a tech demo, but is actually an OS you would want to use on a daily driver device."

Whether the operating system can actually run, and do so with stability, on Apple silicon is up for debate. In a series of posts to Real World Technologies forum earlier this month, creator Linus Torvalds said he would "absolutely love" to own an M1-powered Mac if it ran Linux, but is uncertain such a premise is feasible.

"The main problem with the M1 for me is the GPU and other devices around it, because that's likely what would hold me off using it because it wouldn't have any Linux support unless Apple opens up," Torvalds told ZDNet last week.

Martin agrees that the custom GPU will be a major hurdle, along with effective power management.

"Running Linux on things is easy, but making it work well is hard," Martin says in the Patreon description. "Drivers need to be written for all devices. The driver for the completely custom Apple GPU is the most complicated component, which is necessary to have a good desktop experience. Power management needs to work well too, for your battery life to be reasonable."

Patreon tiers range from $3 to $48 per month, with perks including the ability to vote on the development of upcoming features, patron-only livestreams and more. Martin estimates he needs about $4,000 a month to take on the project, and that goal is 91% complete at the time of this writing.

If the developer takes on the venture, he plans to collaborate with other developers and "anyone else who wants to contribute." A timeline for completion was not discussed.

Comments

  • Reply 1 of 18
    sflocalsflocal Posts: 6,092member
    Considering the huge performance advantages of the M1, and future Apple Silicon, the Linux community will certainly be motivated to get Linux working on Apple Silicon.  If successful, it will leave the WinTel community in the dust.   I can easily see countless Mac Minis being purchased and used as high-performance, power-efficient servers and clusters.  

    I think this is the beginning of the end of the x86(64) dominance for non-mobile computing.  

    edited November 2020 Beatsseanjjeffharriswatto_cobra
  • Reply 2 of 18
    Would love it if they just focused on the headless server aspects first since that would be immediately useful. Linux UI is such an incredibly small niche market compared to servers where linux marketshare is dominant and ARM chips are already amazon instances you can use
    watto_cobra
  • Reply 3 of 18
    anomeanome Posts: 1,533member
    But when will someone port Minix?

    I thought it odd that Torvalds would be so dismissive of the idea, even with the proprietary GPU situation, I shouldn't have thought it would be that difficult. Insane for one developer, sure, but not ridiculous. And the idea that it's not worth the effort seems to be very anti-engineering to me. People have done more difficult things just for the hell of it before, and we could have predicted someone would port Linux to ASi.

    I mean I get that he's trying to take a sideways swipe at Apple over what he sees as a lack of openness, but I think he's grossly underestimated what a bored software engineer will do for kicks.
    watto_cobrarandominternetperson
  • Reply 4 of 18
    wood1208wood1208 Posts: 2,905member
    Matter of the time version of Linux will be running on M1 Macbooks and MAC Pro. Wouldn't be surprised Microsoft will help port Windows on M1 MACs.
    Beats
  • Reply 5 of 18
    BeatsBeats Posts: 3,073member
    Besides price there is no longer a reason to own a Windows virus machine.

    sflocal said:
    Considering the huge performance advantages of the M1, and future Apple Silicon, the Linux community will certainly be motivated to get Linux working on Apple Silicon.  If successful, it will leave the WinTel community in the dust.   I can easily see countless Mac Minis being purchased and used as high-performance, power-efficient servers and clusters.  

    I think this is the beginning of the end of the x86(64) dominance for non-mobile computing.  


    If Apple allows iPhone 12 and above exclusive games your statement would be more true.

    Apple A14 chip has massive potential but Apple caps developers at iPhone 7 or so(I believe). Add an Apple TV M1 and Apple could take over the gaming industry within 2 years.

    EDIT:
    I misread your last sentence as "for non-mobile gaming".
    Will leave my comment up as it makes good points regarding M1 and gaming anyway.
    edited November 2020
  • Reply 6 of 18
    chasmchasm Posts: 3,274member
    Was I hallucinating when Apple appeared to devote a half-minute or so to the emulation powers of the M1, including an on-screen demonstration of some flavour of Linux running just fine on it?

    Sure, that probably wasn't a native version, but if it runs fine in Rosetta 2 I don't see any rush to develop an Apple-native version -- not to mention that the software people tend to want to run in Linux won't be up to speed for a good long time. I'm glad to see another community understanding the great potential for M1 and beyond, but this project seems like an ego trip for "one guy" to take on rather than to organize a team to tackle such a big project.

    I don't doubt Hector's qualifications but this just doesn't seem to be the right approach to solving this particular problem IMO.
    watto_cobraanome
  • Reply 7 of 18
    jdb8167jdb8167 Posts: 626member
    chasm said:
    Was I hallucinating when Apple appeared to devote a half-minute or so to the emulation powers of the M1, including an on-screen demonstration of some flavour of Linux running just fine on it?

    Sure, that probably wasn't a native version, but if it runs fine in Rosetta 2 I don't see any rush to develop an Apple-native version -- not to mention that the software people tend to want to run in Linux won't be up to speed for a good long time. I'm glad to see another community understanding the great potential for M1 and beyond, but this project seems like an ego trip for "one guy" to take on rather than to organize a team to tackle such a big project.

    I don't doubt Hector's qualifications but this just doesn't seem to be the right approach to solving this particular problem IMO.
    There are already several VMs running Arm64 Linux. They work pretty well for being mostly demonstration projects. There is also a pre-beta version of QEmu that can run Arm64 Linux or Windows on Arm. These are not emulated, they are running Arm64 code in a VM on an M1 Mac. 
    edited December 2020 watto_cobrachasm
  • Reply 8 of 18
    swat671swat671 Posts: 150member
    I wonder when Apple will release the macOS Big Sur version of Darwin, and if they'll have both the intel and ARM versions. They have 10.0-10.15.6, including a lot of the drivers. They have both intel and PPC versions where appropriate. So, if/when they make that source code available, I would ASSUME it would make it easier to get Linux and/or other flavors of Unix up on running. 

    https://opensource.apple.com/
    watto_cobra
  • Reply 9 of 18
    mjtomlinmjtomlin Posts: 2,673member
    chasm said:
    Was I hallucinating when Apple appeared to devote a half-minute or so to the emulation powers of the M1, including an on-screen demonstration of some flavour of Linux running just fine on it?

    Sure, that probably wasn't a native version, but if it runs fine in Rosetta 2 I don't see any rush to develop an Apple-native version -- not to mention that the software people tend to want to run in Linux won't be up to speed for a good long time. I'm glad to see another community understanding the great potential for M1 and beyond, but this project seems like an ego trip for "one guy" to take on rather than to organize a team to tackle such a big project.

    I don't doubt Hector's qualifications but this just doesn't seem to be the right approach to solving this particular problem IMO.

    That was a Linux demo was virtualization not emulation, so yes, it was running natively. I think what this project aims to accomplish, is to get an Apple Silicon Mac to boot up under Linux. To do so, you would need many Apple specific drivers, since these systems are almost completely custom.

    Apple was very supportive of Yellow Dog Linux that ran on their PPC Macs, so who knows, maybe they'll work with the Linux community - they mainly seem to target and support Ubuntu Linux in their open source projects.
    watto_cobrachasmbala1234
  • Reply 10 of 18
    dewmedewme Posts: 5,335member
    sflocal said:
    Considering the huge performance advantages of the M1, and future Apple Silicon, the Linux community will certainly be motivated to get Linux working on Apple Silicon.  If successful, it will leave the WinTel community in the dust.   I can easily see countless Mac Minis being purchased and used as high-performance, power-efficient servers and clusters.  

    I think this is the beginning of the end of the x86(64) dominance for non-mobile computing.  

    The M1 definitely exposes one of the inherent weaknesses of the WinTel model. Apple has always had the advantage of controlling both the hardware and the software, but up until now the primary benefit was a better user experience and an overall more refined product. The perceived “cost premium” or “Apple tax” for obtaining these benefits by going with an Apple product somewhat diminished it from obtaining a broad market appeal.  However, with the M1 Apple brings superior performance to the table, and that is definitely going to tip the balance a bit more in Apple’s favor. Paying for performance isn’t an esoteric thing, it’s a bottom line differentiator that resonates with just about everyone. 

    To compete directly against Apple’s M1 every Windows/Linux PC maker would seemingly have to create their own SoC, which is out of the question for all but a few manufacturers. But I don’t think they will even try. Intel, and especially AMD, will most definitely up their game to deliver better performance at lower power levels while still retaining a general purpose compute & graphics engine that allows multiple system vendors to compete. They will also exploit the weaknesses of Apple’s strategies by allowing for greater configurability, more variety of form factors, post sales component upgrades, operating system flexibility, and a greater variety of system configurations at more varied price points.

    The M1 is simply the opening shot in the next round of battles that will take place over the next decade of personal computing. As consumers we will be the primary beneficiaries of the battles that lay ahead. The x86 logjam has finally been broken, and the playing field is clear for a whole new wave of possibilities. 
    watto_cobra
  • Reply 11 of 18
    wizard69wizard69 Posts: 13,377member
    jdb8167 said:
    chasm said:
    Was I hallucinating when Apple appeared to devote a half-minute or so to the emulation powers of the M1, including an on-screen demonstration of some flavour of Linux running just fine on it?

    Sure, that probably wasn't a native version, but if it runs fine in Rosetta 2 I don't see any rush to develop an Apple-native version -- not to mention that the software people tend to want to run in Linux won't be up to speed for a good long time. I'm glad to see another community understanding the great potential for M1 and beyond, but this project seems like an ego trip for "one guy" to take on rather than to organize a team to tackle such a big project.

    I don't doubt Hector's qualifications but this just doesn't seem to be the right approach to solving this particular problem IMO.
    There are already several VMs running Arm64 Linux. They work pretty well for being mostly demonstration projects. There is also a pre-beta version of QEmu that can run Arm64 Linux or Windows on Arm. These are not emulated, they are running Arm64 code in a VM on an M1 Mac. 
    I don't see running ARM code as the big problem.    Rather it is understanding enough about Apples M1 to make use of all the resources on the machine.    Without documentation it will likely take years to decipher everything.  

    I'd love to see Linux on this machine but I really think it will need a change of heart at Apple to accelerate that development.
    watto_cobra
  • Reply 12 of 18
    sflocal said:
    Considering the huge performance advantages of the M1, and future Apple Silicon, the Linux community will certainly be motivated to get Linux working on Apple Silicon.  If successful, it will leave the WinTel community in the dust.   I can easily see countless Mac Minis being purchased and used as high-performance, power-efficient servers and clusters.  

    I think this is the beginning of the end of the x86(64) dominance for non-mobile computing.  

    Reasons why this is incorrect.
    1. Most Windows users actually like it. Just as most Android users actually like Android. So please stop making these endless proclamations on how products that billions of people actually like are going to dry up and blow away in favor of products that generally have from 5% (Macs) to 15% (iPhones) market share.

    2. Even if 1. wasn't the case, what on earth does getting Linux to run on M1 ARM have to do with Windows? You are aware that Linux has been available on ARM for at least a decade, right? And that Linux has been on x86 since ... the beginning? So, what does an operating system that has practically zero end adoption outside the tech world have to do with people switching from Wintel to ... anything?

    3. "I can easily see countless Mac Minis being purchased and used as high-performance, power-efficient servers and clusters."  See 2. High-performance, power-efficient ARM-based servers and clusters already exist and have for 10 years. Mac Minis - which despite all the hype are actually only equivalent to the performance of 6 and 8 core Intel CPUs - offer no particular advantages over the ARM workstations and servers that Ampere, HP, Caldexit etc. already offer. 

    4. "
    I think this is the beginning of the end of the x86(64) dominance for non-mobile computing." Yeah ... Intel - whose 6 core CPUs already roughly equal Apple's 8 core ones - will have 7nm workstation and server chips out 1Q 2021. They are going to have 5nm workstation and server chips out by 1Q 2023 (even if they have to hire TSMC and Samsung to make them). AMD is going to have 5nm desktop and laptop (granted ultrabook/workbook, not notebook) chips out by 4Q 2021. Intel already has a Core i7 NUC device that offers performance similar to the Mac Mini at 10 watts. 

    You guys really do think that Intel's losing Apple's 6% market share is going to crush them? Or that you think that Apple is anywhere close to being able to come up with a chip to match Intel's 56 core Xeon server CPU or AMD's 128 core one? Well hey. come back 3 years from now. We will get a good laugh at these predictions, just like all the "Android will fail!" predictions from 2009-2015 (the iPhone 6 and the Oracle lawsuit was the last, best hope for you guys I guess) are looking pretty hilarious right now.
  • Reply 13 of 18
    Would love it if they just focused on the headless server aspects first since that would be immediately useful. Linux UI is such an incredibly small niche market compared to servers where linux marketshare is dominant and ARM chips are already amazon instances you can use
    Because ... a $699 device that runs an 8 core processor at 5 watts cannot be run as a server. The people who want this are developers who are looking for a Linux desktop development device. Absolutely no one is going to take a device that performs below the level of an Intel Core i7 and has thoroughly inadequate cooling for high performance/high I/O applications and has only 16 GB of RAM to use as a server.

    When Apple products an Mx-based Mac Pro then we can talk about headless workstations and servers, OK? 
  • Reply 14 of 18
    dewme said:
    sflocal said:
    Considering the huge performance advantages of the M1, and future Apple Silicon, the Linux community will certainly be motivated to get Linux working on Apple Silicon.  If successful, it will leave the WinTel community in the dust.   I can easily see countless Mac Minis being purchased and used as high-performance, power-efficient servers and clusters.  

    I think this is the beginning of the end of the x86(64) dominance for non-mobile computing.  

    The M1 definitely exposes one of the inherent weaknesses of the WinTel model. Apple has always had the advantage of controlling both the hardware and the software, but up until now the primary benefit was a better user experience and an overall more refined product. The perceived “cost premium” or “Apple tax” for obtaining these benefits by going with an Apple product somewhat diminished it from obtaining a broad market appeal.  However, with the M1 Apple brings superior performance to the table, and that is definitely going to tip the balance a bit more in Apple’s favor. Paying for performance isn’t an esoteric thing, it’s a bottom line differentiator that resonates with just about everyone. 

    To compete directly against Apple’s M1 every Windows/Linux PC maker would seemingly have to create their own SoC, which is out of the question for all but a few manufacturers. But I don’t think they will even try. Intel, and especially AMD, will most definitely up their game to deliver better performance at lower power levels while still retaining a general purpose compute & graphics engine that allows multiple system vendors to compete. They will also exploit the weaknesses of Apple’s strategies by allowing for greater configurability, more variety of form factors, post sales component upgrades, operating system flexibility, and a greater variety of system configurations at more varied price points.

    The M1 is simply the opening shot in the next round of battles that will take place over the next decade of personal computing. As consumers we will be the primary beneficiaries of the battles that lay ahead. The x86 logjam has finally been broken, and the playing field is clear for a whole new wave of possibilities. 
    1. The WinTel model has had from 80% to 87% market share since 1990. 
    2. The Android-Qualcomm-MediaTek model has enjoyed the same.
    3. Performance is not superior. The Apple M1 only definitely outperforms the dual core and quad core Intel chips that it replaced in the entry level Intel Macs. Intel hexacore chips seem to be a wash and Intel octacore chips outperform the octacore M1 chip. This was based on Apple Insider's own benchmarking. Further, please note that Qualcomm SOCs like the 855 and 865 have long outperformed Intel dual core chips and have recently achieved parity with the quad core ones. The Qualcomm Snapdragon 888 will exceed the performance of all Intel quad core chips and approaches the performance of the hexacore ones. S
    4. Intel and AMD need to "up their game" when an Intel octacore CPU and an AMD octacore CPU already outperforms the M1? Sure, right. Instead, they just need to reduce heat and power. And that is ongoing. Intel 7nm server chips come out 1Q-2Q 2020. AMD 5nm desktop chips come out 4Q 2020. Can we all come back and talk about how "Wintel is doomed!" then?
    5. No battles. AMD will catch up to Apple's 5nm process in 11 months. Intel will go from 14nm to 7nm for their chips that will compete with the M1X (that will be in the iMac, Mac Pro and non-entry level MacBook Pro an Mac Mini) devices by 2020. 

    Look, all PC manufacturers need to do is stop putting quad core CPUs in laptops that cost more than $1000 and desktops that cost more than $700. To the extent that any of them do anyway. The only $1000 laptop with a quad core CPU that I can think of is the Samsung Galaxy Chromebook. And even that was by accident! That Chromebook was originally going to have a Samsung Exynos 992 SOC in it but Google sold them out by not approving Linux beta for it. Samsung would have put a 6-core CPU in it but couldn't because of the heat (it is fanless because again it was supposed to have a Samsung ARM chip in it).

    An Intel Core i7 laptop or desktop provides comparable performance to an M1 Mac. Again, Apple Insider's own benchmarking states this. Most Windows laptops that cost as much as a MacBook Air already have them, or they have a comparable octacore AMD Ryzen 7. 

    As I have been stating ... well since joining, the Apple M1 is this massive development for people who are already devotees of Apple products and services. For the other ... 90-95% of the population? Things will continue to go on as they did. Look, at this point you need to see if M1 Macs can even surpass ChromeOS in market share. Last quarter, Macs had 8%. ChromeOS had 11%. After that, see if M1 Macs can reach 15% market share. Which was the highest macOS market share ever ... though due to consumers and particularly enterprises hating Windows 8 and refusing to buy them. If that happens, great, but expecting more than that will be unfounded.
  • Reply 15 of 18
    jdb8167jdb8167 Posts: 626member
    wizard69 said:
    jdb8167 said:
    chasm said:
    Was I hallucinating when Apple appeared to devote a half-minute or so to the emulation powers of the M1, including an on-screen demonstration of some flavour of Linux running just fine on it?

    Sure, that probably wasn't a native version, but if it runs fine in Rosetta 2 I don't see any rush to develop an Apple-native version -- not to mention that the software people tend to want to run in Linux won't be up to speed for a good long time. I'm glad to see another community understanding the great potential for M1 and beyond, but this project seems like an ego trip for "one guy" to take on rather than to organize a team to tackle such a big project.

    I don't doubt Hector's qualifications but this just doesn't seem to be the right approach to solving this particular problem IMO.
    There are already several VMs running Arm64 Linux. They work pretty well for being mostly demonstration projects. There is also a pre-beta version of QEmu that can run Arm64 Linux or Windows on Arm. These are not emulated, they are running Arm64 code in a VM on an M1 Mac. 
    I don't see running ARM code as the big problem.    Rather it is understanding enough about Apples M1 to make use of all the resources on the machine.    Without documentation it will likely take years to decipher everything.  

    I'd love to see Linux on this machine but I really think it will need a change of heart at Apple to accelerate that development.
    That's why VMs are coming along quickly because you don't need to do that work. The VM virtualizes the hardware. Reverse engineering proprietary hardware is much more difficult. I have no idea if this developer has the chops to do but if he does, more power to him.
  • Reply 16 of 18
    This Cloudguy is pretty funny.  Me thinks he protests too much.
  • Reply 17 of 18
    rezwitsrezwits Posts: 878member
    cloudguy said:
    1. The WinTel model has had from 80% to 87% market share since 1990. 
    And I'll say this HALF short – TWICE strong:

    (VCRs were all the rage from 1970-1995 (25 years), way over 80% market share, haha market share)
    (Disc Players of various types, were all the rage from 1996-2020 (25 years), way over 80% market share, once again, haha market share)
     
    Since 1990, WinTel SLAVES have been saying Apple/Mac is DOOMED.

    Now it's 2020 (30 years later), and Mac FANBOYS are NOW saying WinTel is DOOMED.

    From here on out we are going to consistently SLAM, YELL, ARGUE, or even TROLL, that WinTel is DOOMED.

    Take your own damn medicine now for the next 30 years! and like it, Cloudguy!

    My, my, how times have changed, maybe we'll be wrong, or a...
  • Reply 18 of 18
    MacProMacPro Posts: 19,718member
    rezwits said:
    cloudguy said:
    1. The WinTel model has had from 80% to 87% market share since 1990. 
    And I'll say this HALF short – TWICE strong:

    (VCRs were all the rage from 1970-1995 (25 years), way over 80% market share, haha market share)
    (Disc Players of various types, were all the rage from 1996-2020 (25 years), way over 80% market share, once again, haha market share)
     
    Since 1990, WinTel SLAVES have been saying Apple/Mac is DOOMED.

    Now it's 2020 (30 years later), and Mac FANBOYS are NOW saying WinTel is DOOMED.

    From here on out we are going to consistently SLAM, YELL, ARGUE, or even TROLL, that WinTel is DOOMED.

    Take your own damn medicine now for the next 30 years! and like it, Cloudguy!

    My, my, how times have changed, maybe we'll be wrong, or a...
    The problem with the Intel FanBoys is their inability to 'think different'. I keep seeing test reports that show there is far more going on than is obvious.  Take for example the amount of RAM we 'know' is needed for certain tasks, e.g. number of tracks in Logic Pro X.   Compare the maximum tracks that playback smoothly on an M1 16 GB vs iMac i9 64 GB and it's obvious Intel vs M1 is not a 1:1 comparison ...by a long way.
    rezwits
Sign In or Register to comment.