That’s not actually an example of what I meant, or at least, it wasn’t the way I was thinking of “working”.
I think your system looks really healthy, as a CPU briefly spiking at 190% of 400% available utilization every now and again is not a problem, but suggests everything is appropriately sized, if not a little oversized for this job. Working in my definition would be when the CPU was constantly been driven to 100%, or sustaining 100% then it’s probably underspec’ed.
In your scenario, if the spikes coincide with the start of tracks, that seems to indicate that Roon is fetching the music and may be doing some processing all at the very start, and then it settles back 15% and is may be just pushing packets? Is the bridge function also on the same system as your Core? If this was your bridge, I’d consider it more relevant, but at less than 50% peak utilization, it seems to be cruising to me.
This is Roon Core on an Odroid H2 — somewhat below Roon’s recommended minimum, but evidently quite adequate to the task at hand.
Correct. That’s why the spikes persist longer when fetching from Qobuz than when playing a local file.
Nope. The Roon Bridge(s) are Raspberry Pi’s.
The point that I was making is that, even though I have done nothing special, by way of galvanic isolation, these spikes (which, I emphasize, are present in everyone’s Roon Core at the start of every track you play) are completely inaudible.
I’ve no skin in this game and no interest in spending money for the sake of the latest or coolest thing; I can only report that replacing a cheap network switch (which has my Roon server and NAS hard disk attached) and standard ethernet cable connected to Lumin streamer, with an optical switch, optical cable and a Sonore optical module, didn’t just improve but utterly transformed the sound quality for the better. Happy for someone else to explain.
Endless and useless debate… Or maybe not. One thing is sure, this topic always bubbles heavy emotion to the surface. Some people played with it and experienced SQ difference. Some people tried it and does not experienced any difference. And there are many, who never tried it, but have the loudest voice. Yes, bits are bits, I don’t want to debate this. But based on my experience routers and ethernet cables can make a difference. I don’t know why, I don’t know how, but there are so many not reasonable “tweaks” which can effect SQ. One of my weirdest experience was with two packs of light blue plastic beads. Placed on the right place (behind the speakers, 150 cm high on a shelf) those little beads had an easily audible impact on SQ. Of course I understand why such opinions start heavy reactions. But my philosophy is (more or less) : do not deny anything until you tried it.
Yes, you are right, this is why I wrote “more or less” in that sentence. But I don’t mind to spend lots of time on my hobbies. And in addition, I think there is a difference to deny something, or to neglect it. I don’t want (and I can’t) try everything. And of course there are many things which I deny without spending hours on researching the “truth”. For example flat earth, chemtrail, etc. But listening music and quest for sound quality improvement is something different. Trial and error… all those are time well spent for me.
For me there is a difference. Changing an ethernet cable in my system is quick, easy and not that expensive. But trying to find the edge of the Earth is much more time consuming and expensive. But on the other hand once I would find it would be a fantastic experience to sit there But of course I understand what are you talking about.
I know, I did it on purpose. There are many-many things, which I neglect. This means for me I do not give a s…t on them. But I don’t judge those because I don’t know enough about them. This is the difference. But we are going off topic…
I have Cisco 2960 switches as well. Could you describe how did you solve the optical connection within your system? Which modules do you use? The switch has only one “optical port”, do you use it for “upstream” or “downstream”? Sorry if I asked something stupid, optical connection with Cisco 2960 routers was out of scope for me. Thank you in advance!
You need a Cisco SFP transceiver for that port (plenty used for around $5 or Amazon for $20 - be sure it’s Cisco coded though) and an LC cable. Pretty much plug and play after that. Note that if your endpoint only accepts 10/100 (such as my old Naim Unitiqute) the oM only outputs 1000 so the 10/100 won’t work. The oM should be used before the endpoint. I use a pair and a Cisco switch before the pair, though the switch has to output gigabyte as that’s all the oM will accept (the SFP outs are all gig on 2960’s).
You don’t play off the cable. You play out of local buffer. See my post here for a run down of a test setup and the associated video
I don’t care if a manufacturer can find 1000 people that think the earth is flat.
Instead of spending anywhere from $400 to $1999 on an ‘audiophile’ switch I did all of this for $214:
Cisco WS 2630 48port GBe, 4 port SFP+ 10GBe switch ($65 on eBay)
HP NC 223 Dual SFP+ PCI-E X 8 NIC ($29 on eBay)
Dell R620 dual GBE Copper, dual SFP+ GBe daughter card from Server Monkey for $20.
Three SFP+ SR MM LC 10GBe transceivers and a 20 meter MM OM3 LC/LC fiber cable from FS.com for $100.
So not only did I do this for 1/2 to 1/8th the price of what the snake oil switches go for, I get at least 300% more through put and all the electrical isolation you could ask for.
I’ve tried plenty. The Uptone eR. I even shot a video of the test setup and magically music played with the network cable pulled.
I’ve had WW, AQ, Nordost, Supra. Hell on the AQ piece I shot a video of a $340 cable at 3 feet vs 315 feet of dirt cheap Berktek CAT5e. Somehow the sound stayed the same even when there were no cables attached.
But of course I understand what are you talking about.