Ryzen Embedded systems

Feedback Hardware
1

I’m selecting a Ryzen based industrial miniPC for machine vision use on a mobile robot and would like to experiment with Haiku on it also.

Of the 4 embedded processor series https://www.amd.com/en/products/embedded-ryzen-series do any have traits more likely or not to work with Haiku?

Out of the industrial miniPCs listed here https://www.amd.com/en/products/embedded-minipc-solutions which has the most advanced graphics/processor combo currently (March 2022)
supported by Haiku?

Is there any more advanced (embedded) industrial graphics/processor platform (any type AMD, Intel, RiscV, Nvidia/Arm64) not currently supported, but with interest among Haiku devs?

2

What are you planning to use for motion control ?

3

I have several options for motion control including I2C, CAN bus, USB(serial), or bitbang GPIO. There are a variety of robot bases already built using different control strategies. Haiku need not support them.

Rnet wheelchair controller uses CAN and a reverse engineered protocol https://github.com/redragonx/can2rNET.

A retired FRC(FIRST Robotics Competition) robot uses a RoboRio roboRIO with TalonSRX(CAN) Talon SRX - CTR Electronics

Another uses something similar to hoverboard motors Robotics from Sky's Edge and SimpleFOC compatible motor controllers https://www.simplefoc.com/

ROS2 on Debian Bullseye will be my starting point. Haiku will either be QEMU or installed dual boot if it can. I’m very curious how far I can get in Haiku, but it isn’t a requirement of the project or anything. This is a lazy personal research project, nothing commercial.

I want industrial embedded computers for low power, soldered ram & proc, better capacitor/inductor mounting, wide temp range. Previously used several Rpi up to 400 model, and 4gb Jetson Nano. I need more RAM, so its upgrade time. I think I can get by with 16Gb but 32 is tempting. I would go with a bigger Nvidia Jetson, I like the one I have a lot, but run into lots of libraries that need to be recompiled for Arm64 and that isn’t always successful.

4

Check this sound

5

Not enough RAM in that one. I’ll be loosing the Isaac platform, so will probably be even less memory efficient and I’m already RAM bound at 4gb.
I have the space and power budget for a little bigger unit if V2000 and Vega 11 will work.

6

I’ve started porting machine kit, I’ll get the git repo up soon.

I’d port linux cnc, but it uses gtk and all the good GUIs are in QT now

7

That would be awesome. Thanks.

8

I’ve booted Haiku on v1605B (Vega 8) and v1807b (Vega 11) embedded series, confirmed that it works.

9

Excellent, that is encouraging. Thanks for the specific versions.

10

Anything with a b550 or similar chipset will probably work, I’ll native boot my pc to8ght and report back what the configuration is.

I am getting ready to overhaul 3 large machine centers. I am loathing using linuxcnc and might spend the extra $ for centroid,l oak just to not have to deal with the idiosyncratic problems of gtk wtc.

11

Correct me if I’m wrong, Centroid is free Windows software that can only be used with their hardware controller?
If sufficient code is available to port it to a different OS, chances are they did not open source the part that locks it to their hardware. Which would mean even official hardware wouldn’t work with Haiku unless there is a compatible binary blob. Right?

Looking forward to Machine Kit.

12

Centroid Milling software is entirely windows based currently, there hardware is TCP/IP compatible however.

I am just beggining to work on LinuxCNC and MachineKit, most of the core components seems to be in place.