I’ve got a NAS built in a Node 304 mini itx case that works great, but uses a ton of power. In Unraid (the OS for my NAS) there is some kind of issue with the Ryzen 3900x processor that I’m running that means I have to disable all sleep states - so it’s always at it’s 100W TDP. Power is super expensive where I live so I’d love to find something more power efficient.

Does it make more sense to buy a more recent(ish) 5th gen ryzen in hopes that the sleep states will work, and thus save money by keeping my existing motherboard?

Or I could go with something a bit more interesting. I’ve seen on Aliexpress motherboards with mobile CPU’s soldered which are very power efficient. For example the N100 has an insane 6W TDP and comes on special boards with lots of sata ports and 2.5G networking (link). The worry with the n100 though is that it only officially supports 16G of ram which might not be enough for zfs.

Any thoughts? Is anyone running a power-efficient build who could throw some advice my way? Thanks!

  • nopersonalspace@lemmy.worldOP
    link
    fedilink
    English
    arrow-up
    1
    ·
    10 months ago

    I’ve thought about that before, I’ve used proxmox in the past and liked it. The hope I guess would be that proxmox is better able to handle the physical hardware than Unraid is, and the Unraid can blissfully mismanage it’s vCPUs all it wants! I don’t love the overhead of having a hypervisor, but maybe it would be worth it in this case.

    • Lemmchen@feddit.de
      link
      fedilink
      English
      arrow-up
      1
      ·
      10 months ago

      How much effort have you actually put into trouble shooting the issue? Maybe it is just a wrongly set CPU governor (performance, instead of ondemand or something else)? Or a certain kernel flag that have to be set on boot?

      • nopersonalspace@lemmy.worldOP
        link
        fedilink
        English
        arrow-up
        1
        ·
        10 months ago

        I’ve done a whole bunch of things but the problem is that the issue w/ the OS locking up was intermittent, so really between every change I would have to wait and see and risk downtime.