LinuxCNC on Intel Clear Linux

I have been playing with the "Clear Linux" distro for a while, and I am truly impressed by its blazing fast performance. Clear Linux is an open source project from Intel, and it is highly optimized for Intel platforms, although most of these optimizations also apply to AMD platforms.
From its website clearlinux.org : "Clear Linux OS is an open source, rolling release Linux distribution optimized for performance and security, from the Cloud to the Edge, designed for customization, and manageability."

Clear Linux (CL) works with software packages that are called Bundles. A Bundle is in a way similar to a Flatpak, as it includes all the dependencies and the bundle is thoroughly tested to work on any Clear Linux system. When a package is not available yet as a Bundle, it is also possible to compile software from source, but the recommended way to install a new application is by submitting a package-request, after which the request will be reviewed by the CL team and eventually assigned to a CL resource for creating and maintaining a new bundle for that application.

I have been trying to compile LinuxCNC on Clear Linux, but ran into too many dependency issues that could not easily be resolved as many libraries are not available on Clear Linux yet, and compiling them all one by one is a very time consuming job. As I think that it would be a very interesting case to have a dedicated LinuxCNC bundle created for CL that will allow a "one click" installation of LinuxCNC on Clear Linux, I submitted a package-request to the CL team earlier this week and I am very happy to see that it has already been assigned to an experienced Intel developer who is also active in several CNC communities. I hope this may result in an easy way to install LinuxCNC on Clear Linux and I am very curious to see if any of their performance optimizations may be beneficial to LinuxCNC. I think it could provide a secure and stable way to run LinuxCNC on modern hardware platforms moving forward.

My CL package-request can be found here:  github.com/clearlinux/distribution/issues/3197

Please share your thoughts on this.

So Intel just wasted a lot of resources on reinventing Gentoo, heh.

Sounds like an AI post.

Note I think the term Intelligence is used very loosely.

Securing a cnc computer is extremely easy if you follow 2 steps.

1: physically secure the machine so idiots don’t have access.

2: apart from the network required for connection to a hardware step generator, eg Mesa, don’t connect the machine to a network.

Point 1 is really the first step in general computer security.

First of all, I would like to thank you all for your comments. I hope many more will follow. Please allow me to reflect a bit on your comments and add some additional thoughts.

@NTULINUX: I am not very familiar with Gentoo, but I am certainly curios to see how well Gentoo performs compared to my daily driver LinuxMint 22 with KDE and my new Clear Linux system, which I installed out of curiosity and for evaluation purposes only. It has been a long time since I played around with Gentoo, and back then it was quite a struggle to get things up and running the way I wanted. But that probably applied to most distros back then. Things have changed a lot the past decades... As for Clear Linux, I have seen some recent benchmarks that gave me the impression that the CL team has really achieved substantially better performance compared to most other distros and I am very curious to see if that would also be the case with LinuxCNC running on CL. And yes, I know that perceived user experience and latency are very different things, so let's put it to the test.

@cornholio: I'm not sure what made you think my post was AI generated, but it is not. For now I will just take it as a compliment, as English is not my native language, so seemingly I am doing quite well On the other hand, your remark about "the term intelligence being used very loosely" feels a bit insulting. I hope that is just my poor interpretation and not your intention.
As for the security aspect of CL, I don't care too much about that and I agree with you that the best approach is to keep machine controllers disconnected from general networks. I am primarily interested in the potential suitability of CL from a real time performance / latency perspective. And I don't know if it will be suitable or any better than others, but I am curious....

To all: I am aware that there is a tendency on this forum to be negative about Intel and AMD, and from a CNC latency perspective that is quite understandable and I share many of the frustrations that have been expressed. Newer hardware comes with many more bells and whistles that none of us CNC guys really need. My good old J1900 Mini-ITX boards (GA-J1900N-D3V to be precise) still give me excellent low latency results, and definitely much better than the most recent platforms, but lets face it.... older hardware will eventually phase out and we will have to work with what is available. For most of us that means that using newer Intel and AMD platforms will become inevitable. I have a plethora of computers that I use for my daily work, including a M2 MacBook Pro, some ARM based SBCs and obviously Intel and AMD devices as well, and none of those platforms is perfect. So let's stop bashing certain processor brands, stop pretending that "everything used to be better when I was young" and instead make today's reality work for us as good as possible. Let's also keep in mind that what we are doing with LinuxCNC is trying make general purpose CPUs operate as MCUs, which they are not. So let us not put the blame on Intel and AMD for that. And if one day the newest processors cannot give us the required real time performance that we need, perhaps we may have to move completely away from using general purpose CPUs to do timing critical tasks, which is basically what Mesa cards and similar products are doing for us too. Personally, my ideal system would be an SBC with both A-Cortex application processor cores and M-Cortex MCU cores, so we can move all timing critical tasks to the M cores (which are built for that purpose), and have the general stuff run on the A cores. Then we would create a synergy between processors and software as intended for industrial applications, but that would require a partial redesign of LinuxCNC. Perhaps one day....

Back to Clear Linux and my motivation to bring it up on this forum: When I first learned about Clear Linux I was just as skeptical as most, for obvious reasons. And perhaps Debian, Gentoo or other distros (like CachyOS) may be able to perform just as good. I don't know, but I am curious, so let's find out. My thoughts are that perhaps with the inside knowledge of the Intel team we may be able to turn modern processor platforms into better performing systems for us. So in order to challenge that, and hopefully see some great results, I submitted a package-request. In the worst case situation the CL team will not be able to create a package at all or only create a system that does not perform any better than what we already have. But at least then we know..... Thus far I must say the Clear Linux team seems dedicated to make it work. As a side note: they seem to use the open source Meson Build System, which itself is already interesting and perhaps worth to consider for future LinuxCNC developments, as it can resolve dependencies for different platforms in a much easier way than building software the traditional way. It would provide us with a true cross platform build system and make us less depending on Debian as the preferred distro.
Anyway.... Fingers crossed and let's see....... And if LinuxCNC on Clear Linux does stand out, then we may be able to benefit from that even for those who do not want to use CL. After all, Clear Linux is open source, so anybody who wants to use their performance improvements and implement it in their own preferred distro is totally free to do so. Whatever the outcome may be, personally I think it may be interesting.

I know Tom and I feel you..... I have been working for large corporations long enough to know that everything that started as a passionate creation by totally dedicated engineers, will be transformed into a ruthless money making machine as soon as financially driven managers take over, whose interest is only the quarterly results. And when that happens, ethics are often thrown overboard and customers are fooled by sugar coated stories written by marketeers. Many good companies have been destroyed that way. But I also learned that complaining about it is not going to help and will only result in frustration and negative energy. Hence my suggestion to stop the bashing and try to make the best out of it. Sending out a reminder every now and then about how things suck doesn't hurt, but personally I don't let it prevail, as it does not help.

Ok I reread this.....

So it's not an announcement about your own work, but just a request for someone else to create the packages.

After having had linuxcnc running on Slackware (a good few years ago), Mint runs fine but paying for hosting and such just to support a couple of users wasn't worth it, Debian is the best no fuss route for x86 & RPIs. Tho personally I found the Raspbian image for the 5 a better option, but that was at the end of 2023 and due to the lack of a Linuxcnc SPI driver it was put in a drawer and now runs Octoprint. Debian runs fine and is stable, performance isn't really a big issue when using hardware step generators as the servo thread isn't that hard to dial in with the latency that's needed (mostly it just works), sometimes (depending on your etherent chip, some RT chipsets can be a PITA) ethernet can cause issues, but mostly this is a driver thing that will show up regardless of the Distro you are using . On the other hand, software step generation that always requires a base thread is another kettle of fish, and this isn't related directly to raw clock frequency.
It works,plenty of others use Debian so questions aren't an issue.
Just be prepared if other do try it and have issues you'll be "the guy" that will have to answer the questions. So that'll be via your own experience or conversing with the CL guys.

Most entry level users will want a "plug & go" solution and not have to worry about too much stuff.
The plus with Linuxcnc & Debian is that there is a live version that can be run to see what things are and whether a certain machine looks if it will be usable without installing anything.

I know bein tied to one distro is not the Linux way it is what it is.

As for you find something insulting or not, that's a you issue, you make that decision, pretty much as being offended, that's a you decision. If I was be be insulting I'd either be extremely crass or attempt a Htichenesque type of insult. (Christopher Hitchens btw)

So it's not an announcement about your own work, but just a request for someone else to create the packages.

It is the way the CL team recommends to have applications made available for their distro, by having it officially added to their collection of bundles. Please see the CL website for more information on that. 

Just be prepared if other do try it and have issues you'll be "the guy" that will have to answer the questions. So that'll be via your own experience or conversing with the CL guys.

I understand your concern, and appreciate your warning, but I think you missed the point a bit. Or perhaps I was not clear enough myself. For me this is not about introducing a new platform for LinuxCNC as an alternative to Debian or whatever. It is about curiosity to see if  LinuxCNC would in any way benefit from the advantages that CL claims to bring with their processor tuned and performance optimized distro. If it does, then that would be interesting. The same applies to those, including myself, who have been playing around with RPI, Orange Pi, Arch Linux and other flavors of hardware, operating systems, desktop Environments, LinuxCNC GUIs, etc. Without curiosity the world will never progress. True, for x86/AMD64 platforms we may be best off with Debian and I certainly believe that too, but exploring alternative ways can be interesting nevertheless. 

I know bein tied to one distro is not the Linux way it is what it is.

Well, better one Linux distro than one MS-Windows 

As for you find something insulting or not, that's a you issue, you make that decision, pretty much as being offended, that's a you decision. If I was be be insulting I'd either be extremely crass or attempt a Htichenesque type of insult. (Christopher Hitchens btw)

There is certainly some truth in that statement. But communication and perception is never truly a "you issue". It is a sender, channel and receiver issue, where many factors influence how a transmitted message is being perceived. That is why emails and online posts, including here on this forum, are always a bit tricky and why many forums have a code of conduct. But I guess in this particular case it was mainly me perceiving things incorrectly, for which I apologize. As for Hitchens, a great mind, and not shy to challenge the established views and opinions. I like that