Hello Bees!
I’ve got a couple of projects lined up that I want to use SBCs (single-board computers) for, and I admit that I have very little knowledge about how the different SBCs from different manufacturers compare to each other, so I figured I’d get y’all’s help.
Project 1: Portable media server
This is something I’ve been wanting for a while in order to make long car trips that involve low or no internet access more enjoyable. The basic idea I have is an SBC with a 2-4 M.2 SSDs, wireless, and bluetooth, that I can load up with media and run Jellyfin on, and then connect to with whatever devices I have around (whether that’s a tablet, a smart tv in a hotel, etc). I want to do this as an SBC versus on a laptop partially so I can power it off my car more easily, and potentially have the car play music from it while driving.
I’m leaning towards something like the CM3588 from FriendlyElec is where I’m leaning, so I could RAID 5 some 4TB M.2 SSDs and get ~11.5TB usable (which would match my current Jellyfin home server setup). I’d love to hear if thoughts on this for this kind of portable use case, and any recommendations on alternatives, or other routes to explore.
Project 2: Miniature AI Machine
I’ve enjoyed experimenting with LLMs and StableDiffusion, and I want to make something a little faster and more targeted towards AI without building a 5U GPU server (nor do I have a spare $14.5k for a barebones setup of one). I’ve seen SBCs targeting AI use via baked-in NPUs, or with NPU expansion slots, and I’m interested in what y’all think about this approach.
I’ve also seen people with rPi clusters ostensibly for ML applications, but never any real write-ups on how these perform compared to a regular (E-)ATX machine with a high-end GPU.
I wish I could recommend specific models, but the best I can do is recommend checking the supported devices at the DietPi site as a jumping off point. Good luck!
DietPi is freaking phenomenal! I wish I had more to contribute, but I only have the Pi’s (3b+ and 4 8gb)
Check out !sbcs@lemux.minnix.dev