Except that it has only Roon-essential processes running and a processor chosen not just for raw power but also low electrical noise.
The Nucleus has fewer processes running because it is a ‘lightweight’ Linux based operating system. It’s a lightweight operating system with fewer code in order to lessen the impact any software updates might have on Roon.
Who told you that fewer running processes result in less electrical noise?
I’ll read thru any source document you care to supply.
Processors are not spec’d based on their electrical noise, low or otherwise…
Again, I’ll read any document that speaks of some processors having lower electrical noise.
There are a couple of interesting posts from Danny about the noise of some of the higher spec processors, while reducing what was running certainly made a difference to the SQ from my NAS.
As for the impact of power supplies on digital equipment, try and experiment if you can: you’ll be surprised. I certainly was. Very…
I seriously doubt that the number of running processes is significantly smaller. I know exactly how many processes are running on my Roon Core machine (Ubuntu 19.10). Anybody care to tell me how many running processes there are on their Nucleus? (I thought not.)
In any case, the more relevant statistic is not how many running processes there are, but how many running threads. Roon is pretty heavily multi-threaded.
Even more relevant is total CPU usage. Which is totally dominated by Roon. The OS overhead, in terms of CPU usage is completely negligible (shows up as 0.0%).
Another ridiculous audiophile myth which has absolutely no basis in reality.
Just so we’re not just talking crap, here’s the actual CPU usage on my Roon Core machine.
“netdata” is the program capturing all the nifty real-time usage statistics. A true audiophile would never run such an abomination while listening to music, as it surely creates a horrid amount of electrical noise pollution.
Aside from “netdata”, everything else (“systemd”, “avahi”, “postfix” and “root”) on the system each register as 0.0% which is to say that they total less than 1/20 of 1% of the total CPU capacity (=400%). Roon’s CPU usage is more than 37 times greater than everything else combined.
I can’t find them. Please post a link. Thanks.
I’m sorry, I just don’t have the time to look in detail. I’ve given it a quarter of an hour. Maybe look for his posts on i5, i7 it could have been in an is faster always better discussion. I did quickly find this, though, on processes and components:
Actively running processes besides Roon stuff? It’s about 5. smbd, networking stuff, ntpd. Ubuntu is a pig in comparison.
All of Roon OS is less 50 megabytes… (not gigabytes … megabytes) – it’s ridiculously slim.
No kernel processes? What a miracle!
(More seriously: that was not an answer to the question.)
A clever switch from “running processes” to “space occupied on disk.” Now you’re going to tell me that the space the OS occupies on disk affects the SQ?
I’m telling you nothing affects SQ – We do not advertise Roon OS affects SQ nor have we ever made such a claim. Roon OS is all about the UX. Size does matter when doing an update, which is part of the user experience.
As for “process count”, you seem to be a guy in the know, and should know this does not actually matter – what matters is context switches. If you really need me to list userland processes:
a network manager process (think ifplugd + dhcp + zeroconf + staticip + filesystem watcher)
acpid (idle until you hit power button)
a process to display status on console and watchdog everything
as for kernel, the kernel has been trimmed significantly from what a general OS like ubuntu has… we ship with a small number of modules (thus the limits of device support), and very few subsystems are enabled. Yes, I understand if stuff isn’t used, it isn’t used…but its amazing how much crap is used just because it’s there.
There is no typical Linux distribution server stuff:
cron + friends
systemd + the assload that it comes with
policy kit stuff
If you think your ubuntu is lightweight, you can keep thinking so… but you and I have a very different definition of lightweight. Ubuntu isn’t the fat pig that MacOS and Windows are, but even a minimal Ubuntu is in a different class than Roon OS.
as for 0% CPU, stop looking at your gui tool… instead look at /proc/[pid]/status
additionally we build everything minimally as we can – so while your systemd is spewing to the journals/logs/syslog, Roon OS processes are not.
just check this out:
# cat /proc/1/status | grep ctxt voluntary_ctxt_switches: 1343 nonvoluntary_ctxt_switches: 144 # uptime 17:20:13 up 97 days, 15:55, load average: 0.00, 0.02, 0.00
I’m with you on the SQ point here… but you asked a question about the trimness of Nucleus (Roon OS) and I’m responding to that fact.
There are other things that are awesome about Roon OS – such as the fact that multiple versions of the OS can be installed at the same time (super awesome when it comes to updates), everything running out of RAM so little OS-induced SSD wear, full-reset on reboot, and many more things that make it the best “appliance” we could.
You don’t have the time and I no longer have the interest.
No matter what, enjoy your musIc.
I stand by my original statement:
Roon’s CPU usage is more than 37 times greater than everything else combined .
And that’s without any DSP. Turning on DSP makes the OS overhead even more negligible.
You have me beat on context switches. netdata does realtime monitoring of system statistics multiple times/second. Nothing a true audiophile would consider running.
Great! We’re all agreed. RoonOS is a fine appliance OS.
I was objecting to the assertion that OS overhead in a generic Linux setup (e.g. Ubuntu server) could possibly make a difference in SQ.
Yeah. I can imagine Roon not wanting to get drawn into any such discussions. They are always fruitless and often offensive. I thought you were genuinely interested, otherwise I wouldn’t have wasted my time on you.
Just use your ears
You have your own idiosyncratic way of interpreting what @danny said.
Now, I’m done wasting my time with you.
Ok so are we all still friends , I think I opened a can of worms here with the old SQ question, what can I say I’m a noob!!! It’s more or les 50/50 it does and it doesn’t but the guy who sell the Nucleus @danny that’s a bit of a curve ball and I’d be inclined to go with him, normally manufacturers are the opposite saying you can hear God in their products. Anyways thanks everyone for the input and remember keep on rocking in the free word
Well, I’m not quite saying that. SQ is a funny term – it’s a spectrum of issues, some which I’m sure you can get behind.
I know for fact that some SQ cases are improved. For example, a person that has an Mac Mini that experiences digital breaks in the audio rendering (clicks, pops, breaks) when MacOS’s Spotlight indexing kicks in would definitely see an SQ improvement by going to any Roon OS based device.
I am not claiming is that analog fidelity is improved by some magic of the machine… besides audible noise – the Nucleus is fanless.
Remember, we built this product to solve real problems out there for a large segment of users. Not everyone is deep in the audiophile search for “SQ”. Part of that is the user experience, part of that is to reduce a support burden, and a lot of it is because no one else did it right in a turn-key solution.
I absolutely agree. Spotlight is a real resource hog and if it’s grinding away in the background, you will experience slowdowns…
That’s one of the reasons why — despite being a dyed-in-the-wool mac person — I’d be reluctant to recommend a Mac Mini for Roon Core.
Better something running a variant of Linux, whether that be a Nucleus, a NUC running ROCK, or RoonServer atop some other Linux variant. The musical experience from any of those will be identical.
Hi @Jacques_Distler any link to where I could pick up a NUC with ROCK pre installed or mabe its not to difficult to install at home.