When the Orion O6 mini-ITX motherboard with a CIX P1 12-core Armv9 SoC was announced in December 2024, we were told binary releases would start on January 15, 2025, and the source code would be released later in Q1 2025. Tom Cubie further explained that upstreaming of the CIX kernel would start in Q2 of 2025, and Linux 6.1 with Device Tree plus Linux 6.6 with ACPI would be supported in the meantime.
I’ve been able to quickly check the Debian 12 binary release on the Orion O6 motherboard at the end of January, but as I was looking for an updated image for the second part of the review, I noticed there weren’t any new Debian 12 images so far. That does not mean there haven’t been any work done, as the forums are somewhat active and I’ve been told Radxa is even more active on Discord, although I and others don’t feel it’s the best place to communicate due to the ephemeral nature of chat rooms…
However, one piece of good news is that Radxa released the initial source code for the Orion O6 board just a few days ago, and the following is available on GitLab:
- cix_opensource – GCC toolchain, drivers for GPU, ISP, VPU, NPU, Linux 6.1.44, etc…
- cix_private – ACPICA tools and utilities designed to assist developers and engineers in working with the ACPI implementation
- linux_repo – Some build scripts and file structure for Debian 12 rootfs
- tools – Some binaries. Mostly rootfs and tools for I2C, SPI, etc… found in /usr/share/cix/bin/
But this SDK release is not ideal. Most importantly, it’s not the code used by Radxa, and just an unmodified vendor SDK. Radxa’s code is scheduled to be released later this week. Although due to their agreement with CIX, some repositories and binary assets are omitted until CIX gives them a greenlight. There are also a few known issues related to video playback, GPU performance depending on the DisplayPort output used, etc… Those will be fixed in the March 2025 monthly release, I assume at the end of the month.
As a side note, while Radxa dumped the SDK on GitLab, they plan to keep on doing the development on GitHub because users from China, Hong Kong, and Macau can’t use GitLab for free anymore:
I think we will stick with GitHub. GitLab is only for SDK dump.
GitLab is pretty hostile to users from China. They will close accounts if they believe you are from China to push you to their Chinese partner, which has turned off the free tier entirely. The international version also introduced member limits before, which we were negatively impacted when we have one such group to communicate with our customer. The overall experience is pretty bad. I can not recommend their product anymore.
Mainline Linux upstreaming has also started with the first commit in LKML by CIX adding CD8180 and CP8180 variants of the P1 SoC and initial support for Orion O6.
If we look further down in the thread, we learn more about the expected vendor tree source code and upstreaming schedule (message from February 27, 2025):
… Radxa will open our v6.1 device tree vendor kernel source code at the beginning of next month. For ACPI vendor kernel source code based on v6.6, it has planned to open at Q2 of this year.
For upstream, we will submit support based on device tree first, since some common code changes are needed for supporting ACPI well at Arm, it may involve lots of discussion. After most of IP drivers are supported by upstream kernel, we will start ACPI upstream work.
There will also be a website to track the Linux upstreaming process, but it’s not up yet or I missed it. That’s all good news, but that also means you’ll need to be patient if you want a board that works with random ISO image compliant with UEFI and/or ACPI. Right now, making the board work to your needs may require lots of reading and tweaking.
For the second part of Radxa O6 review, I think I’ll test the current images with Debian 12 and Linux 6.1 to run a few benchmarks and see what works and what doesn’t, and then compile Linux 6.13, and try again with the same tests. That’s unless a new image comes up within this week of course.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.
Support CNX Software! Donate via cryptocurrencies, become a Patron on Patreon, or purchase goods on Amazon or Aliexpress
My board, ordered on 31-dec-2024, from ALLNET china, still hasn’t shipped.
there’s a thread in the forums about this. It seems that allnet hasn’t shipped anything yet while arace has started. But I might be making broad generalizations out of a few samples though.
I can confirm that Arace has started shipping. Ordered my board on Feb 2nd, shipped on Feb 18th and received Feb 24th. All of this to Europe 🙂
I just got a reply from ALLNET saying that shipping will be delayed by 2-3 weeks due to higher than expected demand and chips shortage.
Radxa has been working with Free/Net/Open BSD to get the system booting there: https://marc.info/?t=173726182600001&r=1&w=2.
As a result I’ve now got a booting OpenBSD system using ACPI boot and a standard openbsd install image and can use any of the packages. Have not really put much on the system yet as I’ve been out
I’ve rebuilt the bios, but honestly it looks like there are still many blobs in it. I’m not sure at all about the relevance of the rare sources that are touched, since when you modify them it has no impact on the final one (e.g. I tried to see if it was possible to get CPUs to work at their advertised frequencies).
The code provides interesting information about the SoC’s capabilities. We learn that it supports DDR5-6400 (hence the claim of “>100GB/s” in public docs), and that the max frequencies of the little cores is apparently supposed to be 2.0 GHz (though I’ve lost that one).