I don’t recall seeing idle power draw for NUCs running ROCK (or for the Elac Discovery, which is ARM if I’m not mistaken). It’d be interesting to get data on that.
If it can make you feel better, it’s probably rather small when idling, which is most of the time (these guys say 3-7w for a NUCi7 running Windows, with an average around 6w). That’s not too bad.
Another option to reduce power consumption, if you use other stuff (like a NAS for your storage), may very well be a converged appliance running Roon and the other stuff.
UnRaid works rather well for this, and their architecture has the huge advantage, from a power-usage perspective, of allowing you to keep parity on data drives without making them into a RAID array. You pay for this in performance, of course, but the way their topology works, you’re only writing to storage + 1 or 2 parity drives during write, and only hitting a single storage drive (and an SSD storing the docker + library data) during read, so you’re not spinning up an entire RAID array each time you want to get your Celine Dion or Jay-Z fix.
Assign a few of your cores to the Roon docker if you’re concerned other stuff going on might impact SQ, and you’ve pretty much virtualized ROCK (which may also be a possibility in a setup of the sort).
Hello again Brian. You were very gracious with your time in this thread regarding running the Core on ARM SoC devices. You were clear regarding their inadequacies at the current time.
Could I however turn you attention to the current crop or Ryzen Embedded SoC offerings? Could you please have a look at the article and board linked to from that and evaluate if you believe it would be sufficient to run the core? Based on Passmark scores of the SoC used, it would seem so. Would love your opinion:
The CPU is probably fast enough. The eMMC would be an area of concern–I have never seen one that behaves like a “real” SSD under load.
$378 is a lot of money for what you’re getting. You could assemble a NUC8i3, SSD, and RAM For less, and the NUC would slightly outperform. Only benefit of the AMD is TDP.
You would run Linux on it, it’s amd64 arch so our linux builds will run. No harm in trying if that’s interesting for you.
Thanx for the comments. Noted and TBD on the eMMC.
But regarding cost. Two or three years ago my 7th Gen i5 NUC I built my RoonCore with, the cost of entry of the NUC plus RAM was higher than this device, and I bought it all for the cheapest I could find; of course one does get the complete package in the lovely NUC case, plus I did have the ability to use a M.2 nVME for the OS and Roon, which is major. But more importantly when I went to purchase a NUC recently I discovered the prices were a good deal higher now than then.
With your comments in mind it does seem as a penny-pinching RoonCore solution this is pretty high on the value curve. I thank you again for them. I will [eventually, not immediately tho] pick this or similar up to see if it is in fact a sufficiently performative platform for the task of a RoonCore.
As far as faster than RPi (or maybe a better performant ARM system might be a better descriptor), I have 2x (soon to have 2 more) LX2160A class nodes in 1u sitting at home. 64GB RAM, NVMe, 100GE, and 16 A72 cores. If that wouldn’t suit a core deployment, not sure what would
Maybe for a small library size. The other arm request will never happen unless the market is like 10% of this kind of deployment. Way too specific imho.
There’s the chip – and then there’s the OS. iPadOS is still very much a mobile OS, focused on (energy) efficiency and battery saving – and then there’s sleep and memory management. There’s no way a persistent database-driven application like Roon Server would not be killed by the OS within minutes.
The iPad is a locked down ecosystem with built-in app closure which is why the Roon client keeps getting shut out of memory for consuming too much battery. Imagine that with the full server environment.
Plus you need a full version of Dotnet which doesn’t exist on iPad os.
It’s good to dream so dream on
iPads are all about power saving and shutting down background processes. This is quite the opposite of what a server OS need to do. They always have 10+ hours battery life for a good reason…no matter how powerful the chips inside get.
I would welcome an arm release, even for those of us who would be prepared to beta test these scenarios. With Ubuntu supporting .net how much work would be involved to release a beta version?
Although the latest 8 core arm solutions like the Rock5, Vim4 aren’t quite up to NUC x86 Gen 6-11 speeds, they do have great multicore results and arguably keep up with my gen 4 i5 NUC which has no issue running Roon (Performing procedural tasks, etc)
I find these two boards very interesting candidates:
Just wondering if the Roon team would chime in here.
With Roon developing native Apple silicon Roon versions, would the Roon team consider releasing an ARM version for general beta testing, that could run on Ashai Linux OS?
I’m thinking,
Apple M1/M2 + Ashai Linux = Super fast/silent/low power Rock Server.
