Mac Mini as Roon Core?

Function ROON
#1

Hey all, I’m considering in investing in a Rose RS520. I’ve demo’d it before and like the looks, build quality, large screen, sound quality, but… don’t love the Rose App. I will use mostly Tidal for streaming (and want to use it outside of the Rose app), and realize there’s no Tidal connect (yet…?). This brings me to Roon, which would also let me build my offline library. Seems like an OK trade off to forget about Tidal Connect, but start using Roon.

I was wondering if I could use a Mac Mini M2 as Roon Core instead of a dedicated Roon device like the Roon Nucleus ONE. Any thoughts or experiences? Pros / Cons? Could then also use it for other home related things (IoT, running a Plex Server, Movie Library, connect to NAS, etc.)

@BorisM - I’ve read that you’re the Roon expert here. Hope OK to tag you. Curious to get your input.

#2

cakepilot,

I’m not the Roon expert but I have been using the new Mac Mini as a Roon core and it doesn’t even break a sweat. It’s rock solid (no pun intended).

StandardModel

1 Like
#3

:slight_smile:

Would work perfectly fine, even a NAS box with an in all likelihood quite a bit slower CPU than an M2 has no problem running Roon, Plex, Surveillance Station with a bunch of cameras and a few other things.

As long as you have some reasonable amount of RAM and aren’t trying to edit 4K video while upsamling to DSD512 it will work well.

Note that with some recent versions of Mac OS there were issues with the built-in firewall forgetting that Roon is a trusted app, but it seems that Apple had fixed that.

1 Like
#4

You think I could run all of this this on the NAS itself? My plan was to run a Mac Mini in headless with Roon, Plex, some IoT stuff, and have my library on a NAS connected to the Mini. A bit pricier, but thought that would be more performant.

#5

Depends on a NAS. I am using Synology DS-1522+ and everything generally works fine. Doing DSP on DSD256+ is questionable (really, the PCM to DSD conversion after DSP) but everything else seems to be fine. QNAP has models with an even faster processors.

Mini M2 is definitely a faster box… but how large is your Roon library? Unless you are pushing above 100K tracks, if you have some other use for a somewhat more expensive NAS, might as well run Roon on it. If anything, if you ever run out of performance on it, you can always migrate to that Mac.

1 Like
#6

Makes sense. Appreciate the input

#7

You’re welcome!

As a general note, while you could most likely hack it to run on any x64 based NAS (as long as you can put a VM or a docker container on it), officially only Asustor, Synology, and QNAP x64 models are supported.

Roon’s recommendation is to have the Roon database (not the actual music files) on an SSD drive. You could waste a slot on that, but at least with some Synology models (QNAP has something similar) you can add a couple m2 drives that can be configured as either a storage volume or as general cache. I would suggest doing the latter, and having the database on a regular spinning disks volume – with anything reasonable that you can run on a NAS, all of the database will fit in the cache (in effect, it will be on SSD) and you will get a speed-up for all other random access.

Roon can have performance issues that seem to be mostly occurring if you have lots of tracks (i.e. bootlegs or something) that can not be identified in any of Roon’s metadata sources. For me, I don’t think I’ve ever had to restart it outside of either Roon or NAS firmware updates, and it all works fine.

#8

To be honest, having an excuse to buy a little Mac Mini wouldn’t be all that terrible, even if in theory everything could run on a NAS :sweat_smile:. I’ve read that running the DSP — as we chatted about in the other thread — may use quite some compute, so would avoid a bottleneck with the M2 Mini. Plus, then I can also goof around with running LLMs locally :slight_smile:

#9

True, although unless you are messing with DSD, even DSP does not particularly tax the processor.

Does the Mini have a GPU good enough for LLMs?

#10

Not for training larger models, but for running pre-trained ones or maybe training smaller ones with LMstudio should be sufficient.