ODROID H3 and H3+

@soeffi yes, I’m also very impressed.

Note that I measured power consumption of H3 device only - it’s DC measurement after power supply (AC/DC converter).
It’s not measured on AC side (~230V) where is not only H3 device but also power supply which has some losses (efficiency about 80-90% usually).

With newest BIOS 1.11 and OS Debian 11 + DietPi + kernel 6.0.0-0.deb11.6-amd64 when nothing is running:

see details about my running system on H3 - idle:

my HW configuration:
» RAM: 1x 16GB Crucial 3200MHz (CT16G4SFRA32A)

  • but it works on lower frequency because Odroid H3(+) doesn’t support 3200 MHz RAM
    (it supports up to 2933 MT/s (DDR4-PC23400))

» SSD: 1x SATA 480GB KINGSTON (SA400S37480G)

  • in my experience and measurements m.2 drives are a bit more power hungry

Some OS / SW details:

  • powertop report: PowerTOP report (using command powertop --html=report.html)

  • my whole dmesg log: dmesg@odroidH3_1.11_6.0.0-0.deb11.6-amd64 - Pastebin.com

  • neofetch:

     neofetch 
           _,met$$$$$gg.          dietpi@saf 
        ,g$$$$$$$$$$$$$$$P.       ---------- 
      ,g$$P"     """Y$$.".        OS: Debian GNU/Linux 11 (bullseye) x86_64 
     ,$$P'              `$$$.     Host: ODROID-H3 1.0 
    ',$$P       ,ggs.     `$$b:   Kernel: 6.0.0-0.deb11.6-amd64 
    `d$$'     ,$P"'   .    $$$    Uptime: 2 days, 4 hours, 23 mins 
     $$P      d$'     ,    $$P    Packages: 280 (dpkg) 
     $$:      $$.   -    ,d$$'    Shell: bash 5.1.4 
     $$;      Y$b._   _,d$P'      Terminal: /dev/pts/0 
     Y$$.    `.`"Y$$$$P"'         CPU: Intel Celeron N5105 (4) @ 2.900GHz 
     `$$b      "-.__              Memory: 172MiB / 15849MiB 
      `Y$$
       `Y$$.                                              
         `$$b.                                            
           `Y$$b.
              `"Y$b._
                  `"""
    
  • htop screenshot:

  • dietpi-benchmark (two runs):

    Benchmarks completed: (max 12.95 W ⚡️)
     - CPU performance : Duration = 2.15 seconds (lower is faster)
     - CPU temps       : Idle = 36 °C | Full load = 42 °C         
     - RootFS I/O      : Write = 312 MiB/s | Read = 444 MiB/s     
     - RAM I/O         : Write = 1807 MiB/s | Read = 4837 MiB/s 
     
    Benchmarks completed: (max 12.90 W ⚡️)
     - CPU performance : Duration = 2.12 seconds (lower is faster)
     - CPU temps       : Idle = 32 °C | Full load = 38 °C
     - RootFS I/O      : Write = 310 MiB/s | Read = 443 MiB/s
     - RAM I/O         : Write = 1771 MiB/s | Read = 4851 MiB/s
    
    • dietpi-benchmark on Raspberry Pi 4 (2GB RAM) (with SATA SSD over USB3) for comparison:
    RPi4:
    Benchmarks completed:
     - CPU performance : Duration = 7.66 seconds (lower is faster)
     - CPU temps       : Idle = 40 °C | Full load = 45 °C
     - RootFs I/O      : Write = 28 MiB/s | Read = 214 MiB/s
     - RAM I/O         : Write = 344 MiB/s | Read = 681 MiB/s
    

I have idle power consumption around only 1.1 — 1.2 W in average ! see: Imgur: The magic of the Internet

In order to achieve such low power consumption we need apply a couple of tweaks:

  • upgrade BIOS to newest (version 1.11 in this time - end of January 2023)
  • run: powertop --auto-tune (before that, install it: sudo apt install powertop)
  • in BIOS: enable ASPM (I enabled also SA GV but only -0.1W, maybe it was measurement error)
    • without ASPM idle power consumption: 2.6 — 2.8 W :zap:
  • make sure in idle CPU core frequencies are 800MHz (using powertop tool)
  • to decrease power consumption even further you can use eMMC module instead of SSD. But it’s most expensive storage (considering price per GB) https://www.hardkernel.com/shop/128gb-emmc-module-h2-2/ see forum: H3 - how to reduce power consumption ? - ODROID
    • If you really want to have a low power machine, forget SSDs and use eMMC instead.


//edit: later (in one week) I will add measurements with some running docker containers… I plan to migrate from RPi4 to H3. I have some services running in docker: portainer, unifi-controller, pihole, plex and Home Assistant. I would like to run HA Supervised, but I’m not sure if it’s possible on DietPi. There is built-in support for Home Assistant in DietPi - but it is installed within a dedicated Python environment. I don’t know much about it.