Open Source Computing

[mrbeastie0x19]

Now I know FreeBSD is the best open source operating system (no citation needed) but I wanted to spark a discussion around the future of Open Source Computing, particularly on consumer devices.

On phones Android and IOS is king. Android uses the Linux kernel. It is supposed to be an open-source platform, but can any of you find the source code? Yeah, they don't make it easy. IOS is obviously not open source, even though some components (the XNU kernel) are.

On desktops Windows and Mac OS X are still king. Neither of which are open source.

For servers and embedded devices Linux is doing quite well (perhaps not as good as FreeBSD which boasts the PS4 and Netflix ). That's one plus for open source software, and one market which is looking good.

The real issue though (imo) is the lack of Open Source Hardware. Even the system I am running now is on some proprietary Intel chip that has another proprietary Intel chip running a closed source version of Minix on it (Intel ME). Same with phones, devices are often using binary blobs for drivers. We might have Linux or FreeBSD but the general state of the hardware ecosystem is an absolute dive. The BIOS is also often proprietary. Device drivers are proprietary. Chip software is proprietary.

I'd really like to see some investment in RISC-V, perhaps a kind of 'Linux' approach to the hardware world where device manufacturers all produce open source specifications which make money by competing and providing support services.

 

[D-FENS]

Now I know FreeBSD is the best open source operating system (no citation needed) but I wanted to spark a discussion around the future of Open Source Computing, particularly on consumer devices.

On phones Android and IOS is king. Android uses the Linux kernel. It is supposed to be an open-source platform, but can any of you find the source code? Yeah, they don't make it easy. IOS is obviously not open source, even though some components (the XNU kernel) are.

On desktops Windows and Mac OS X are still king. Neither of which are open source.

For servers and embedded devices Linux is doing quite well (perhaps not as good as FreeBSD which boasts the PS4 and Netflix ). That's one plus for open source software, and one market which is looking good.

The real issue though (imo) is the lack of Open Source Hardware. Even the system I am running now is on some proprietary Intel chip that has another proprietary Intel chip running a closed source version of Minix on it (Intel ME). Same with phones, devices are often using binary blobs for drivers. We might have Linux or FreeBSD but the general state of the hardware ecosystem is an absolute dive. The BIOS is also often proprietary. Device drivers are proprietary. Chip software is proprietary.

I'd really like to see some investment in RISC-V, perhaps a kind of 'Linux' approach to the hardware world where device manufacturers all produce open source specifications which make money by competing and providing support services.

Android is a fork of the Linux kernel, it does not use Linux. It is heavily tweaked and stuffed with phone, tablet and other mobile device drivers some (or maybe most??) of which are not available in Linux.
Furthermore it is infested with Google's spyware programs that leak all your files into their cloud.

Can you find the source - sure, the source to the kernel is here: https://android.googlesource.com/kernel/common/
There is a lot of proprietary software around the kernel which is not opensource so you don't get to read the code.
They forked the Android source into the open source project Cyanogenmod, which is now called LineageOS and you can install that replacing your stock Android. LineageOS is mostly open source but it still depends on the firmware blobs for each device (because the manufacturers don't reveal the code, there is no other way).
My personal opinion as a LineageOS user - it is a nice try, but in practise it sucks. It's hacky, only geeks would do it. Updating is a major hassle. And all interesting Apps are on Google Play only, so you are kind of stuck with a very limited set of apps.

What we need is not a better copy of Android but an actual Linux (or maybe FreeBSD ) actually running on mobile devices. For that we need a committed phone manufacturer and also a sensible GUI shell (KDE is making an effort, also Ubuntu) and many more ports of existing Mobile Apps to Linux.
We've come a long way in this regard, but there is still no sensible option on the market AFAIK.
Worth watching though: https://tuxphones.com/list-linux-mobile-devices/

Desktops - the market moves steadily towards mobile. Soon desktop will not matter. Even today you have a lot of jobs that can be done on a smartphone and you don't need a computer. There will be a market, but it will steadily shrink.

Regarding hardware - it has a high barrier to entry. And the manufacturers prefer keeping things proprietary, unless there is some kind of direct benefit in opening up. As the mass consumer does not care about openness, logically we don't see any change in this direction. I would not hold my breath.

 

[mrbeastie0x19]

Android is a fork of the Linux kernel, it does not use Linux. It is heavily tweaked and stuffed with phone, tablet and other mobile device drivers some (or maybe most??) of which are not available in Linux.
Furthermore it is infested with Google's spyware programs that leak all your files into their cloud.

Can you find the source - sure, the source to the kernel is here: https://android.googlesource.com/kernel/common/
There is a lot of proprietary software around the kernel which is not opensource so you don't get to read the code.
They forked the Android source into the open source project Cyanogenmod, which is now called LineageOS and you can install that replacing your stock Android. LineageOS is mostly open source but it still depends on the firmware blobs for each device (because the manufacturers don't reveal the code, there is no other way).
My personal opinion as a LineageOS user - it is a nice try, but in practise it sucks. It's hacky, only geeks would do it. Updating is a major hassle. And all interesting Apps are on Google Play only, so you are kind of stuck with a very limited set of apps.

What we need is not a better copy of Android but an actual Linux (or maybe FreeBSD ) actually running on mobile devices. For that we need a committed phone manufacturer and also a sensible GUI shell (KDE is making an effort, also Ubuntu) and many more ports of existing Mobile Apps to Linux.
We've come a long way in this regard, but there is still no sensible option on the market AFAIK.
Worth watching though: https://tuxphones.com/list-linux-mobile-devices/

Desktops - the market moves steadily towards mobile. Soon desktop will not matter. Even today you have a lot of jobs that can be done on a smartphone and you don't need a computer. There will be a market, but it will steadily shrink.

Regarding hardware - it has a high barrier to entry. And the manufacturers prefer keeping things proprietary, unless there is some kind of direct benefit in opening up. As the mass consumer does not care about openness, logically we don't see any change in this direction. I would not hold my breath.

Interesting I did not know about LineageOS, I did know about Google's 'googlesource' but it seems so unintuitive compared to git, if I want to get the FreeBSD or Linux kernel sources I can simply download a tarbell, it looks to me like Google make it deliberately hard to download all components (Is there a way to git clone the whole repository of Android code?) And yes as you say a big issue is the ecosystem.

I think Ubuntu on phone is interesting and yes I agree we need a serious alternative to Android. I do also agree with what you say about the shrinking desktop, but desktops are still much more powerful and will continue to be used because of games, graphics software, office suites and development tools.

And sadly you are very correct with hardware. Entry is very high. Not only in cost but in knowledge, I hope they open up more in the future, or pressure is put on them to do so. I think there are projects to flash or reverse engineer the ROM shipped with some of these components (not sure about the microcode? I think not), same with reverse engineering device drivers (but then the issue is legally using open source binaries without manufacturer consent). Whatever the situation, I do hope in the future it changes significantly. RISC-V is the most promising I have seen for ISAs and could set a precedent for other manufacturers if it becomes popular.

 

[bsduck]

And all interesting Apps are on Google Play only, so you are kind of stuck with a very limited set of apps.

You can get what's on the Play Store without a Google account (except for paid apps) using https://auroraoss.com/download/AuroraStore/

 

[D-FENS]

You can get what's on the Play Store without a Google account (except for paid apps) using https://auroraoss.com/download/AuroraStore/

This does not work in practise. I think once I was able to install something, but all the other times I actually needed it, it simply did not work anonymously.

 

[bsduck]

Well, it works for me, but I have to admit I don't use many apps.

 

[D-FENS]

Interesting I did not know about LineageOS, I did know about Google's 'googlesource' but it seems so unintuitive compared to git, if I want to get the FreeBSD or Linux kernel sources I can simply download a tarbell, it looks to me like Google make it deliberately hard to download all components (Is there a way to git clone the whole repository of Android code?) And yes as you say a big issue is the ecosystem.

I think Ubuntu on phone is interesting and yes I agree we need a serious alternative to Android. I do also agree with what you say about the shrinking desktop, but desktops are still much more powerful and will continue to be used because of games, graphics software, office suites and development tools.

And sadly you are very correct with hardware. Entry is very high. Not only in cost but in knowledge, I hope they open up more in the future, or pressure is put on them to do so. I think there are projects to flash or reverse engineer the ROM shipped with some of these components (not sure about the microcode? I think not), same with reverse engineering device drivers (but then the issue is legally using open source binaries without manufacturer consent). Whatever the situation, I do hope in the future it changes significantly. RISC-V is the most promising I have seen for ISAs and could set a precedent for other manufacturers if it becomes popular.

Indeed, it is not easy to compile and tweak Android. I tried compiling LineageOS from scratch (which includes Android) but I gave up after git cloning more than 80 GB of stuff (and counting), I said - this is madness. I have no time for this crap.
And I do compile my Linux kernel at every update, it takes only 2-3 GB and about 3 minutes to compile.

Point 2 - you have games on phones and people started gaming only on phones for a while. Of course PC-s will be around, just not as often as they did 10 years ago. Office stuff is also doable on a phone, emailing (people use whatsapp and weechat nowadays btw.)

Point 3 - reverse engineering a device for the purpose of making an open source driver for it, is it legal? This is completely jurisdiction dependent, so check with your lawyer. I think the US and Europe allow this, as long as you don't copy any original material but create the program from scratch (at least what I read in the Reversing book by Eldad Eilam), but don't quote me on this.

If you purchase a device the manufacturer should not be allowed to block you from putting your own software on it in my opinion. If you listen to Richard Stallman, he would tell you simply not to buy those devices. Period.

 

[mer]

hardware is, in a word, hard.
lots of upfront investment, potentially long time to market, then you need people to buy the hardware before you see any return on the investment.
That's why lots of hardware manufacturers make you sign NDAs and other legal stuff to get the documentation you need to write a proper device driver. Yes, I know "not all of them do", hence my "...lots of..." phrasing.
Can you do a lot without the documentation? Sure, depends on what the device actually is.
CPUs: once they say the family, you have a starting point.
Lots of other devices will run in a compatibility mode of some kind (compatible with previous generations of device) but you need the docs to figure out the new stuff.

Phones, Laptops, there have been attempts to create open versions of the hardware in the past. Not sure what is still viable.

Toss in current global supply chain concerns and one winds up needing lots of money to even think about getting started.

my opinions.

 

[mrbeastie0x19]

Indeed, it is not easy to compile and tweak Android. I tried compiling LineageOS from scratch (which includes Android) but I gave up after git cloning more than 80 GB of stuff (and counting), I said - this is madness. I have no time for this crap.
And I do compile my Linux kernel at every update, it takes only 2-3 GB and about 3 minutes to compile.

Point 2 - you have games on phones and people started gaming only on phones for a while. Of course PC-s will be around, just not as often as they did 10 years ago. Office stuff is also doable on a phone, emailing (people use whatsapp and weechat nowadays btw.)

Point 3 - reverse engineering a device for the purpose of making an open source driver for it, is it legal? This is completely jurisdiction dependent, so check with your lawyer. I think the US and Europe allow this, as long as you don't copy any original material but create the program from scratch (at least what I read in the Reversing book by Eldad Eilam), but don't quote me on this.

If you purchase a device the manufacturer should not be allowed to block you from putting your own software on it in my opinion. If you listen to Richard Stallman, he would tell you simply not to buy those devices. Period.

You know I may have been a little short sighted here on Point 2. One of my key points was that desktop systems will continue because of faster typing, you could in theory just attach a keyboard or mouse to a phone, and if you wanted a bigger screen cast it to a TV. Nonetheless better processing will mean they continue to be used for a long time for things like software development and games, although as you say the games market is quite big on phone.

 

[bsduck]

People using a computer for whatever kind of productive work, be it office work, multimedia processing, programming, etc. can't do that efficiently on a phone or a tablet. That's quite a lot of people, not just a few geeks. Sure, most people may use a mobile device rather than a PC to read the news and send messages, but that won't make desktops and laptops disappear or become a niche anytime soon.

 

[mrbeastie0x19]

People using a computer for whatever kind of productive work, be it office work, multimedia processing, programming, etc. can't do that efficiently on a phone or a tablet. That's quite a lot of people, not just a few geeks. Sure, most people may use a mobile device rather than a PC to read the news and send messages, but that won't make desktops and laptops disappear or become a niche anytime soon.

That was my original point too, but I definitely think advancements in mobile tech could offset this, and as I just said you can attach a mouse, keyboard, or even external monitor, I could see it becoming a viable option at some point for even sophisticated applications. But rn the desktop has a very clear advantage for these things.

 

[bsduck]

That's not only a matter of hardware, but also of software... interfaces designed for touch screens are nothing like efficient in a typical desktop scenario.

 

[mrbeastie0x19]

That's not only a matter of hardware, but also of software... interfaces designed for touch screens are nothing like efficient in a typical desktop scenario.

They work fine for tablet devices, which people do use for graphics (I know designers that use the Ipad anyway).

 

[mer]

That's not only a matter of hardware, but also of software... interfaces designed for touch screens are nothing like efficient in a typical desktop scenario.

This is one of the trends that I absolutely detest: "everything must have a touch screen interface".
Sorry, there are times when old fashioned knobs and buttons are best.
Example: automobiles. putting in touch screens to control everything. You want to adjust the climate control: a simple knob to adjust the heat, a button to press to turn on A/C, buttons for fan speed. But no, all that is now in a fancy touch screen that forces you to take your eyes off the road because you can't simply feel yourself push a button.
Toasters: really? I need an app to control it?

It's great when used appropriately, horrible when forced.

 

[Geezer]

Open Source Hardware

Not sure that hardware has a 'source' as does compiled code.

Now I know FreeBSD is the best open source operating system (no citation needed) but I wanted to spark a discussion around the future of Open Source Computing, particularly on consumer devices.

On phones Android and IOS is king. Android uses the Linux kernel. It is supposed to be an open-source platform, but can any of you find the source code? Yeah, they don't make it easy. IOS is obviously not open source, even though some components (the XNU kernel) are.

On desktops Windows and Mac OS X are still king. Neither of which are open source.

For servers and embedded devices Linux is doing quite well (perhaps not as good as FreeBSD which boasts the PS4 and Netflix ). That's one plus for open source software, and one market which is looking good.

The real issue though (imo) is the lack of Open Source Hardware. Even the system I am running now is on some proprietary Intel chip that has another proprietary Intel chip running a closed source version of Minix on it (Intel ME). Same with phones, devices are often using binary blobs for drivers. We might have Linux or FreeBSD but the general state of the hardware ecosystem is an absolute dive. The BIOS is also often proprietary. Device drivers are proprietary. Chip software is proprietary.

I'd really like to see some investment in RISC-V, perhaps a kind of 'Linux' approach to the hardware world where device manufacturers all produce open source specifications which make money by competing and providing support services.

'Open Source' is a band waggon.

'BSD License' might mean something.

 

[mrbeastie0x19]

Not sure that hardware has a 'source' as does compiled code.



'Open Source' is a band waggon.

'BSD License' might mean something.

There's multiple factors to it. First is the process. Second is the design such as non proprietary cpu designs (arm vs risc-v). Third is the software you can find in ROM that's flashed onto the device itself, here you find things like the bios, device firmware, microcode updates etc, these have software components as well as hardware but are usually treated as part of hardware.

 

[kpedersen]

This is one of the trends that I absolutely detest: "everything must have a touch screen interface".

I agree. I am still more productive, even on a touch screen device when the interface is mouse oriented.

Touch screen interfaces are too big and basic with not enough information or functionality on each screen. They are also fullscreen rather than smaller windows making multi-tasking slow and awkward.

It is basically like being back on MS-DOS but with no real keyboard.

 

[Sevendogsbsd]

I have always been a big proponent of open source computing and for me, I have never had a hardware issue. I only use open source on PCs though - never on a laptop so perhaps that is why I have never had issues. Also, I custom build my PCs so I have control over the parts.

My chief complaint about open source, specifically the graphical "desktops" is the number of choices. There are too many. I am not saying choice is a bad thing but from the eyes of a consumer, too many choices in this realm is a bad thing. Think about Linux and the number of "distros". Random consumer says " I want to try Linux". Which one? Over 100 to choose from. Then, choose a desktop. Which one? I am not saying this deters people from using open source operating systems but it would certainly be overwhelming for the uninitiated. Most consumers think of computers and phones as commodities. They don't really "care" about them as long as they work. Using an open source operating system means you view using a computer as more of a passion than the average consumer. In my opinion, this is why open source desktops have never really gotten a foothold.

As for phones, people want something they can absolutely count on and I am not sure the open source offerings are there yet, at least in terms of mass consumption.

Don't get me started on touch screens, I hate them, with the exception of phones. I do not own a tablet and have not owned a tablet since the Nexus 7, which I gave to my wife and she used briefly until it became unbearable because it was so slow. For me, a keyboard and mouse are the way to go for productivity. That and I can do anything on my phone or laptop that I can do on a tablet so what is the point of a tablet? One more thing to charge....

 

[Jose]

I'd really like to see some investment in RISC-V, perhaps a kind of 'Linux' approach to the hardware world where device manufacturers all produce open source specifications which make money by competing and providing support services.

There's also Openpower. Raptor Systems markets a completely open workstation:

OpenPOWER - RCS Wiki

wiki.raptorcs.com

Note that these two examples are the extremes of the hardware spectrum. RISC-V plays at the small low-power end, and Openpower plays at the workstation end. This is exactly where you would expect changes to happen. The middle is a brutal space dominated by high volumes and low margins.

This is one of the trends that I absolutely detest: "everything must have a touch screen interface".

Don't get me started. I'm fanatical about a clean computer screen. This one place I worked, people would come over and wanted to point things out on my screen with their fingers. I got tired of handing them pens, so I actually bought and old-school pointer that I would hand to them. I'm sure they're still making jokes about that to this day, but I don't care. My screen was fingerprint-free. The idea of touching your screen on purpose sends cold shivers down my spine.

 

[mer]

Jose your last paragraph? Me too. I would act like a Catholic school nun with a ruler when people tried to do that to me.

 

[ralphbsz]

Now I know FreeBSD is the best open source operating system ...

Depends on how you measure "best". If you measure it by democratic means, namely market acceptance, then it is very very bad, with a market share that is somewhere around a percent, or much less.

Android uses the Linux kernel. It is supposed to be an open-source platform, but can any of you find the source code?

Yes, easily. Try googling for "android linux kernel source".

For servers and embedded devices Linux is doing quite well (perhaps not as good as FreeBSD which boasts the PS4 and Netflix ).

For servers, Linux is unbeatable. On supercomputers (the Top500 list for example), it has 100% market share. That's not a joke: There is no single supercomputer in the 500 fastest machines in the world that uses any operating system other than Linux. Nearly all of the hyperscalers (the FAANG and friends) run on Linux, with the exception of Microsoft, which has a lot of servers running Windows. In commercial usage (banks, insurance, health-care), Linux is the lion's share, Windows is the rest, and all the others (the *BSDs, plus the proprietary Unixes like AIX, HP-UX, Sun's OSes, and zOS) are all at a percent or two, sometimes less.

Even looking at Netflix: The bulk of Netflix runs on AWS, and the a lot of uses Linux. There are some parts that use FreeBSD (the CDN), others that don't.

From a market share viewpoint, Linux has completely won the war on servers. On handhelds and desktops (which includes laptops), the story is different. And the embedded market is complicated.

But there is one thing that one needs to understand: The open source arena is dominated by professionals. Most open source contributions are made by people whose full-time job it is to write the software. Today, hobbyists are a very small fraction of open source. Fundamentally, the Linux ecosystem has become a way for a few dozen large companies (led by RedHat, now a division of IBM) to cooperatively develop software and share the cost.

The real issue though (imo) is the lack of Open Source Hardware. ... I'd really like to see some investment in RISC-V, perhaps a kind of 'Linux' approach to the hardware world ...

No problem. If you have a few billion, you can invest it in open source hardware, and seriously compete with Intel. That's for example what AMD did, in their rebirth a few years ago: People invested a few billion in AMD, and AMD managed to get a market share of about 10-30% (depending on how you measure). I bet with a few hundred billion, you could actually create a serious competitor. Are you offering your savings account?

Sorry, but the amount of engineering required to create high-end chips (both the chip itself, and the fab and the integration between fabbing and chip design) is so huge, it is out of the realm of hobbyists, and small companies. And the chip industry has not seen fit to use their (very large) investments to create a cooperative environment.

Why not? Here's my speculation: About 30 years ago, computer companies discovered that you can't make much money selling operating systems (with the exception of MicroSoft, which was actually making good money with Windows licenses). The people selling MVS, VMS, Primos, GCOS, AIX, HP-UX, Solaris, SysVr4... were not making significant money off the operating system itself. Sometimes they made money by bundling (you get VMS for cheap when you spend millions on a VAX, or you get MVS for cheap if you by a 308x or 309x). Then Linux came, and within 5 years the bottom dropped out of the OS market, with the exception of Windows. That's why they were willing to give up control of the OS source code, since writing OSes is a (commercially) hopeless thing anyway.

But processor chips are different. Both Intel and AMD make an enormous amount of money by selling you CPUs. And don't get me wrong: they're not a ripoff. Those CPUs are really good, and they are also a good value: the people who need computers get enough economic benefit out of the computers that giving Intel or AMD about $1000 for a chip is a reasonable investment. The Intel/AMD duopoly was also such a good value that it has nearly eliminated all competition (I think PowerPC is the last server CPU left standing). In the last few years, the rise of Arm has changed the equation somewhat, in particular in the handheld/laptop market, but those high-end server Arm CPUs that the likes of Amazon and Google are using these days are not orders of magnitude cheaper than the equivalent Intel/AMD CPUs. So, big companies make big money selling CPUs. And making good CPUs is very investment intensive. A few decades ago, a handful of chip designers could build a whole microprocessor in a few months (the first 8-bit and 16-bit micros were created by small engineering teams); today doing a mainstream CPU takes a cast of thousands of highly trained and specialized people. Given that investment in IP and the expected profits, it seems implausible that companies will simply share it.

 

[mrbeastie0x19]

Depends on how you measure "best". If you measure it by democratic means, namely market acceptance, then it is very very bad, with a market share that is somewhere around a percent, or much less.


Yes, easily. Try googling for "android linux kernel source".


For servers, Linux is unbeatable. On supercomputers (the Top500 list for example), it has 100% market share. That's not a joke: There is no single supercomputer in the 500 fastest machines in the world that uses any operating system other than Linux. Nearly all of the hyperscalers (the FAANG and friends) run on Linux, with the exception of Microsoft, which has a lot of servers running Windows. In commercial usage (banks, insurance, health-care), Linux is the lion's share, Windows is the rest, and all the others (the *BSDs, plus the proprietary Unixes like AIX, HP-UX, Sun's OSes, and zOS) are all at a percent or two, sometimes less.

Even looking at Netflix: The bulk of Netflix runs on AWS, and the a lot of uses Linux. There are some parts that use FreeBSD (the CDN), others that don't.

From a market share viewpoint, Linux has completely won the war on servers. On handhelds and desktops (which includes laptops), the story is different. And the embedded market is complicated.

But there is one thing that one needs to understand: The open source arena is dominated by professionals. Most open source contributions are made by people whose full-time job it is to write the software. Today, hobbyists are a very small fraction of open source. Fundamentally, the Linux ecosystem has become a way for a few dozen large companies (led by RedHat, now a division of IBM) to cooperatively develop software and share the cost.


No problem. If you have a few billion, you can invest it in open source hardware, and seriously compete with Intel. That's for example what AMD did, in their rebirth a few years ago: People invested a few billion in AMD, and AMD managed to get a market share of about 10-30% (depending on how you measure). I bet with a few hundred billion, you could actually create a serious competitor. Are you offering your savings account?

Sorry, but the amount of engineering required to create high-end chips (both the chip itself, and the fab and the integration between fabbing and chip design) is so huge, it is out of the realm of hobbyists, and small companies. And the chip industry has not seen fit to use their (very large) investments to create a cooperative environment.

Why not? Here's my speculation: About 30 years ago, computer companies discovered that you can't make much money selling operating systems (with the exception of MicroSoft, which was actually making good money with Windows licenses). The people selling MVS, VMS, Primos, GCOS, AIX, HP-UX, Solaris, SysVr4... were not making significant money off the operating system itself. Sometimes they made money by bundling (you get VMS for cheap when you spend millions on a VAX, or you get MVS for cheap if you by a 308x or 309x). Then Linux came, and within 5 years the bottom dropped out of the OS market, with the exception of Windows. That's why they were willing to give up control of the OS source code, since writing OSes is a (commercially) hopeless thing anyway.

But processor chips are different. Both Intel and AMD make an enormous amount of money by selling you CPUs. And don't get me wrong: they're not a ripoff. Those CPUs are really good, and they are also a good value: the people who need computers get enough economic benefit out of the computers that giving Intel or AMD about $1000 for a chip is a reasonable investment. The Intel/AMD duopoly was also such a good value that it has nearly eliminated all competition (I think PowerPC is the last server CPU left standing). In the last few years, the rise of Arm has changed the equation somewhat, in particular in the handheld/laptop market, but those high-end server Arm CPUs that the likes of Amazon and Google are using these days are not orders of magnitude cheaper than the equivalent Intel/AMD CPUs. So, big companies make big money selling CPUs. And making good CPUs is very investment intensive. A few decades ago, a handful of chip designers could build a whole microprocessor in a few months (the first 8-bit and 16-bit micros were created by small engineering teams); today doing a mainstream CPU takes a cast of thousands of highly trained and specialized people. Given that investment in IP and the expected profits, it seems implausible that companies will simply share it.

Trust me I understand that the semiconductor business is outside the possibility of all but the richest and most clever, but so were compilers and kernels at one point, and things change a great deal, we can flash rom so in a way we don't even need to have truly open hardware, as long as we can change the software but I think kernel developers would greatly appreciate more open hardware cooperation. And yes in many ways Linux is by far the most used in the modern world counting all those devices out there. Still a man can dream! I do believe that hardware should be open, huge barriers aside. Despite the huge barriers cpu architecture and microarchitecture have changed considerably and continue to do so, I wouldn't say that the current players will never be replaced, and maybe one day they'll be replaced by a company that has more interest in openness.

*Also android is far more than the kernel and you know this, go download the whole thing and compile it and tell me that was an easy process compared to compiling freebsd.

 

[Vull]

Depends on how you measure "best". If you measure it by democratic means, namely market acceptance, then it is very very bad, with a market share that is somewhere around a percent, or much less.
...

I measure "best" as "fastest and most reliable" at deploying Apache, PHP, and PostgreSQL." Here FreeBSD wins by substantial margins. I don't care about popularity contests.

 

[mark_j]

ralphbsz
The thing is, all the companies you mentioned associated with these OS, like Prime, IBM, HP, Dec, Sun etc all produced their own chip(s) and hardware. They all sell/sold "enterprise" hardware and the software was tied to it. You couldn't run Solaris (SunOS) on a non-Sparc, or Primos on a non-Prime cpu. It was never about the OS software, hell some gave it away, like Prime or for a small fee (compared to the hardware), OpenVMS. It was always about the hardware. You'd buy a Pyramid for north of $1million and they'd throw in Unix free. HP sold off OpenVMS and now vmssoftware is trying to port it to Intel because it was so tied to the dead Itanium and Alpha.

But, back on topic, mrbeastie0x19, yes, your best chance is with RISC V. The problem is the very small amount of hardware available is catastrophically high compared to cheap ARMs and yet just as able.

 

[mark_j]

Don't get me started. I'm fanatical about a clean computer screen. This one place I worked, people would come over and wanted to point things out on my screen with their fingers. I got tired of handing them pens, so I actually bought and old-school pointer that I would hand to them. I'm sure they're still making jokes about that to this day, but I don't care. My screen was fingerprint-free. The idea of touching your screen on purpose sends cold shivers down my spine.

In these days of covid, it would be triply so. ?