Do I really need an " Audio Grade Network Switch "?

UpTone EtherREGEN with AfterDark clock if you're serious about digital.

No, you do not. The real key to audiophile level performance is the streamer. Then when you get to the level of performance you are looking for you can start tweaking  interconnects, power cords, and maybe add a EtherRegen.. (~$600).

I run my system (see my ID) using a $69 wifi extender. My system achieved the highest level of performance equivalent to my really good vinyl end  using just that. I have since added a EtherRegen and received a tiny increase in SQ. I think it is because I have a really good streamer which cashes the data and completely isolates the outgoing stream from the network. 

It is a good question...

I recently implemented an sotm sNH-10G switch with a FARAD3 PSU and a AQ Hurricane Source PC and after more than one week I am happy happy with:

- the increased space and timing

- a more open presentation

But I believe a lot will depend on ones DAC...

@antigrunge2 are you saying that the clock in a switch makes a difference?

In a home network, the clocks in a switch will make absolutely no difference, even if one of the switches’ clock has shitted there is no implication on the frame ordering because the synchronization happens at the physical layer by means of Ethernet synchronization messaging channel more specifically in the 3bit SSM field. Once this is stablish and negotiated, then you have the Adaptive Clock that adjusts based on the receive buffers.

If you want more information about how timing affects networks ITU-T G.8261 and IEEE 802.3ay are a good place to start.

In no way or shape, a switch running as a switch, and not performing some higher layer operations, will touch the the data.

If you are running a home network with hardware that meets the Ethernet specs, there isn’t anything you can do to improve timing, especially if it is wired. In the case wireless you could do a number of things in the signal but then this is also regulated and specified.

Even if a switch could provide 1^3 precision, the additional timing stamp will be dropped, not even rounded, because there isn’t space in the SSM field to accommodate higher precision; and the reason why there isn’t an accommodation for higher precision it is because it is not needed in any use case to date.

as @ghdprentice correctly points out it is the streamer work, after the NIC has forwarded the data to the upper stack, to fill its receive buffers and then dispose of them to the processor in a precise and timely manner so the music plays in a perfectly timed way. I have not studied the hardware stack on a streamer, but I would be hard pressed to believe that a streamer needs to be more time accurate than direct memory transfers among cluster nodes. If streamers and DACs need more time accurate than what Ethernet can provide they would be using a complete different protocol, and network adapters. Maybe IB, but then since they are receiving the data from an Ethernet network, there would be no more timing precision.

Maybe if the streamer had an IB network adapter and was connected to the the streaming server via IB, you could benefit for a more precise reference clock, and even then it is not the network’s work on timing how data gets processed in the receiving hardware.

Although I know that I over simplifying , all a network needs is common and negotiated clock good enough to deliver packets in as an orderly manner as possible, it is the NICs to reassemble, if needed, the frames and packets it receives and move them to the upper stack.

I hope my English on this is good enough to explain.

Jawn! I listen and know what I hear. There are numerous views on other fora and they seem to HEAR what I hear. Also: in a USB connection between streamer and dac the server’s clock gets slaved to the dac’s clock and thereby becomes irrelevant.