Home Assistant Community

What cpu are you using in your NUC?

#21

Running a Intel NUC 8 [(NUC8i3BEH) - Core i3, Tall] and it has zero problem running home assistant in a docker container. I also have containers running for deconz, mariadb, grafana, influxdb, portainer and several other services unrelated to home assistant. It uses virtually no resources.

It’s overkill for home assistant, but I like the small form factor and that I can run other things on it like the network software for my ubiquiti network products.

1 Like
#22

the intel cpu and gpu are integrated, so you’ll get the same graphics unit as in a nuc.

1 Like
#23

I’ve just setup a Chinese NUC clone with an I5 8250, 120g ssd and 8gb Ddr4. Running 4 cameras though it, sits at 30% load @51degC. Fan barely spins.
Pretty happy with it so far. So much faster than the rpi3 I had before.
Running hass.IO direct from an ssd. (Etched the image direct to the ssd via a usb caddy.)

8gen is worth the premium over the 6/7 gen in terms of processing SPEED, power use and Ddr4 support. Chinese clones are about 60% the cost compared to the Intel versions so worth it. Go with a reputable seller in Ali and should be no risk. Wait for a sale and good cash bask and should come to $350aud.

1 Like
#24

So you put the NUC Hassio image directly on the Chinese PC’s SSD rather than using the docker install method?

Did you have any issues?

#25

I used my own ssd, but yeah, just used the sd card install method for the rpi3, but used a usb sata case to house the ssd. Works great.
Would be abit harder with an m.2 drive. But yeah all good so far.

1 Like
#26

Just installed NUC version on HP Z220S, very zippy compared to PI 3+.

Writing to USB stick, then inserting in target system expecting a write to internal disk/ssd, does NOT work.

Using Etcher to write to (future) internal disk/ssd and then install disk/ssd in target system does work !

#27

i’m using a Intel Skull Canyon NUC with 16GB and very happy … in one year i went from RPI3/HASSIO ->RPI3/Debian->Intel NUC/Ubuntu16->Intel Skull Canyon NUC/Ubuntu18 … i’m now using 25 video cameras, tensorflow on 5, 5 HADashboards … don’t underestimate how much horsepower you will need … i did

#28

Got the NUC image on my Chinese PC. All looks good.

Did you figure out a CPU temperature sensor for the Chinese PC?

#29

For mine, it show under Glances.
‘iwlwifi 1’

1 Like
#30

Is that really the CPU temp,or something to do with wifi?

#31

Intel N4100 + 8go RAM
I have Proxmox hypervisor with HA + docker in a VM, and some other VM/container like OMV, Plex, aircast …

If you have only HA to run, a RPI is enough but if you want more a mini pc is a must, and the fact you have HA in a VM with Proxmox allow you to backup and restore the whole VM in 1 click, very useful and much more convenient and reliable than copying an SD card, or make a snapshot in Hassio.

FYI I have a Kodlix GN41 but I don’t recommend it as it is very cheap : touchy to open because plastic is very cheap and the fanless cooling is very bad.

The best mini pc I recommend to start with are :

  • Gigabyte GB-BLCE-4105
  • Gigabyte GB-BLPD-5005
  • Intel NUC7PJYH
  • Zotac CI329 Nano
#32

I just switched over from RPi3 to NUC8i3, 8GB RAM, 480 GB SSD and it’s a huge boost when fiddeling around with config and rebooting but eventhough when switchen through views and rendering lovelace sensor cards.

I run Ubuntu Server on it and installed Hass.io for generic Linux on it.

My setup runs with an average load of 2%, so plenty of ressources available for other projects.

#33

I have my hassio installed on my nuc on a m.2 drive…no problems

#34

I no nothing about Proxmox other than it looks like an opensource vm. What formats does it support?

By that I mean I was originally going to try and run hass.io under my synlogy that uses another open source vm. The problem was they only had images for VMWARE so it was not compatible (There was a difficult work around but I went the pi route to get up and going).

Can you run hass.io directly under Proxmox or does it have to be Proxmox>Linux>hassio?

#35

Also, if you compare CPU’s and you know the model number i usually take a peek at https://www.cpubenchmark.net/high_end_cpus.html and see if it has a higher number or not. Of course this does not take into account things like Quick Video Sync support on intels. On the latest models this lets you encode/decode 4k video with barely noticable effect. Plex users want this normally :slight_smile:
And if you have something with Nvidia you could use that the boost your tensorflow speeds i believe.

#36

Is it possible to monitor system (Intel NUC) with Glances if I have Proxmox as virtualizer?

I would like to monitor cpu temperature that is not available with System Monitor Component

Thanks

#37

I’ve got a question for all of you… I’ve been contemplating about moving to a NUC and using docker but how do you guys manage all the ports?
I think PiHole for example likes port 80/443, then NGINX needs these two ports as well.
Would I be able to run both of these docker containers and just change the ports?
I saw no difference with PiHole running the web interface in any other port, but not using port 443 makes makes website load slower. And about NGINX I haven’t tried other ports as well, so I wonder how feasible it is. Having 2 raspberry pi 3b+ and an Odroid XU4 certainly helped me a bit as they’re separate hosts, also if one goes down the remaining keep running. So Hassio goes down but I still have Unifi or PiHole, but the reliability it is directly tied to these boards and the damned microSD cards which will fail sooner or later…

I want to run NGINX, PiHole, Cloud9, DuckDNS, MariaDB, Unifi, PLEX, Nextcloud and also use the same machine as a NAS. How feasible is it?
I was thinking of OpenMediaVault on a NUC and then running all of the aforementioned apps in docker on top of it. And what kind of hardware would be better? Right now I have an Odroid XU4 running PiHole, MySQL, Nextcloud and PLEX (I can barely do a single 1080p transcode but it’s just for me) and it is taking roughly 15% on idle and 25% RAM… and a Raspberry Pi 3B+ running the rest (sitting at 10% CPU, 80% RAM).

So how feasible to run the above items? What hardware should I look at? I was thinking on i3, 8GB RAM, 128GB SSD.

Something that concerns me is also the ability to run Hassio on docker as I like the add-ons feature…

As of now I tried running normal Homeassistant on docker and it was a complete failure, no Unifi or NGINX proxy manager docker would work because there’s no MySQL 5.6.x (or MariaDB 10.x.x) or something like that for ARM which is required for NGINX proxy manager (official docker). I’m fairly new to Linux stuff and docker.
I found docker amazing but many of the Dockers available are not a one click solution and you must provide JSON with external dependencies such as databases. I’d really like to run Hassio on Docker…
Also I’d like to have BLE working as I have a few MiFlora BLE sensors.

#38

Went with NUC i3 8th gen. Future proof and very low power usage ~4.3watts at idle

#39

I use I microserver gen8.
Running windows server essentials with Hassio on Ubuntu VM.
I have found that this is by far the best low power always on system available for home use.
Currently runs:-
Blue Iris
Plex
Home assistant
File server

#40

J5005 NUC7PJYH, Quad Core Pentium, runs very well and great BFYB.

I use Ubuntu 16.04 LTS, and use Docker to run;

  • Home Assistant
  • Motion Eye
  • Open VPN
  • Portainer
  • TasmoAdmin
  • Plex
  • Unifi

Everything runs very well, no issues.