Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
' first' open-source ARMv9 motherboard surfaces – Radxa Orion O6 (tomshardware.com)
45 points by rbanffy on Dec 20, 2024 | hide | past | favorite | 24 comments


Inauspiciously but unsurprisingly, both the "Downloads" and "Documentation" tabs on the product page for this "open-source" board are empty.

Some of their other boards have PDF printouts of the Cadence OrCAD schematics, and even DXFs of the top and bottom layers of the board, but none have actual hardware source documents.

Their Github repositories contain little more than pre-built images for the boards.

Older boards do have CPU block diagrams and pinouts of the IO ports (which should be standardized, but it's nice to have them in one place) but there are no real datasheets for any of these CPUs.

Yes, they claim "Everything about Radxa SBCs is open" but I don't see it.

Ah, now I see how you'd customize one of these open-source products: Their "Services" page offers "Personalized PCB logo and customer SKU, and changes to pin-compatible materials, such as Wi-Fi speed or expanded operating temperature range" with no non-recurring engineering costs if your order is more than 500 pieces or $50,000" or you can call them for custom base-plate design or other mechanical changes to the layout.


…and they cost more than an x86 based SBC or even a mini PC.


I pre-ordered one of these mostly because it exposes dual 5 Gbps Ethernet and PCIe Gen 4... so the expansion capabilities and I/O are the best of any Arm system I've seen short of $1,500+ Ampere systems.

But the marketing bar graphs and lack of efficiency data, along with not (yet) seeing any of the board firmware (says it uses EDKII) makes me a little worried.

I've been burned buying new Radxa hardware in the past, without ever seeing software support materialize later (or half-implemented features that only work on one or two barely-maintained distros...).

But I would love to have another option for an efficient Arm board. Hopefully the efficiency is also close to M1 which is pretty amazing.


I was going to tell you to wait for Jeff Geerling to do this first otherwise you’re in for a world of hurt.

Then I read your username. Big fan :)


So it was true :-)

https://news.ycombinator.com/item?id=41885705

Really looking forward to your review of this.


Radxa did pretty well with their RK3588-based boards, but we'll see if this gets the same treatment. I really want to preorder one, but I'm also highly skeptical about software support. I don't care about "open source" firmware or anything, I just want to be able to boot mainline Linux and use all of the advertised features.


I really want to buy something like this but then you find out there is no mainline kernel support so you need to use their modified Ubuntu kernel and while not impossible you are on your own with anything custom, including firmware and boot code.


It's only open source before you populate it. After that it has a ton of closed-source chips on it. Including the ARM CPU that's the whole point of it existing in the first place.


Isn't that their work, the design of the board? The ARM isn't theirs, they can't open source it. That's valuable. I'm guessing that anyone that wants to design another ARM board can leverage this work.


Can't see the source


It seems marketing BS… will they open source the ARM710 Verilog/VHDL?

Probably the board schematic… what a joke


By that logic any OSHW project using integrated circuits is closed-source. After all, the PDK is proprietary, the fab's production pipeline isn't publicly available, and there isn't even a way to confirm the HDL supplied is actually what is being used in the chip!

If we want truly open-source hardware we'll have to go back to relay computers. Everything beyond that is "a joke".


No love for a verilog design in an fpga as open source hardware?

It’s a flawed surrogate of course - lower performance and higher cost than an asic produced at volume - but the open source hardware aspect is very strong.


The circuitry (ie transistor level) of the FPGA is still closed no?


Yes, but there are many different fpga’s, many of which many verilog will run on, which to my mind makes the downsides of proprietyness low enough that this might be the sweet spot to stop at going down the stack..

however awesome an open source fabbed asic or fpga would be of course.


It's not open unless you can make it from sand.

Actually, it's not open unless you can make it from hydrogen atoms.


That’s going too far. RYF certification is what you could reasonably require.


> By that logic any OSHW project using integrated circuits is closed-source.

There are degrees to things. It's not cool to try to claim the prestige of "open source" when the entire board is nothing but a vehicle for your very-much-closed processor design.


> when the entire board is nothing but a vehicle

You can get the board design and order it manufactured yourself, or you can change it to suit a different processor that's more open.


What about the many real open source RISC V?


Do we know anything about the Cix people, whose core this is? I've never heard of them before!

I'm really glad to see ARM finally after decades have something semi competitive versus buying an x86 mini-pc.

One question, hoping the x16 slot has bifurcation support. But this seems like such a perfect board for a SSD NAS. If the slot can be divided into x4 or ideally down to x2 lanes. I'm not sure what bifurcation support entails, what it requires: anyone know?


A CD8180 on a raspberry pi form factor would be nice


RYF certification or no sale.


does it support SBSA or anything?




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: