Page 2 of 2

Re: MOTU 16a Thunderbolt vs AVB to Mac

PostPosted: Sun Dec 16, 2018 12:51 pm
by jkinkennon
I just took a look at Dante and the first thing I run into is the virtual soundcard with latencies of 4, 6, or 10ms. Tell me this isn't the best we can expect. I understand that Dante is a fine product and would be useful in many situations, but for a PC or Mac to multiple speakers via ethernet there's no advantage to Dante or AVB if we end up with USB latencies. I think I'll look for info on Mac latencies using AVB and see if there is some definitive info.

Re: MOTU 16a Thunderbolt vs AVB to Mac

PostPosted: Sun Dec 16, 2018 1:19 pm
by murph
Hi John,
dedicated Dante cards have far lower latencies. The same goes for AVB-AVB direct connections. (Both around 0.6ms)(AES50 can go lower, but that's another discussion....)

It's the computer buffering that causes the latencies in virtual card (and Apple AVB), as the main processor is doing the work and sharing it's resources with everything else.......

Tony.

Re: MOTU 16a Thunderbolt vs AVB to Mac

PostPosted: Sun Dec 16, 2018 1:50 pm
by rhedgebeth
We have not noticed Dante latencies to be perceptible with the virtual soundcard implementation in usual circumstances. That being said we will be using a hardware version, with sub millisecond latency, in an upcoming concert hall installation. In that situation we're looking at an acoustical latency of some 40 milliseconds from the speaker location to the front of the stage and we don't wish to add to that - even a little. I believe the accepted threshold of latency perception is generally regarded to be about 12 milliseconds.

Re: MOTU 16a Thunderbolt vs AVB to Mac

PostPosted: Sun Dec 16, 2018 5:57 pm
by dw154515
jkinkennon wrote:I just took a look at Dante and the first thing I run into is the virtual soundcard with latencies of 4, 6, or 10ms. Tell me this isn't the best we can expect. I understand that Dante is a fine product and would be useful in many situations, but for a PC or Mac to multiple speakers via ethernet there's no advantage to Dante or AVB if we end up with USB latencies. I think I'll look for info on Mac latencies using AVB and see if there is some definitive info.


I did a Dante system recently across an university's Enterprise network (mentioned on another forum thread around here somewhere) where from endpoint to endpoint, through several switches, about 1/2 mile from transmitter to furthest receiver, we still had <0.5ms latency.

Re: MOTU 16a Thunderbolt vs AVB to Mac

PostPosted: Sun Dec 16, 2018 9:13 pm
by jkinkennon
It sounds as if we are confusing network latency with the latency including the Mac or PC with their drivers. I did see a $1000 PCIe card which would replace the virtual sound card. Is that part of the sub-1ms setup? Either AVB-TSN or Dante can work wonders across their networks, but we need to include the computer and drivers.

Since 10ms latency is the equivalent of moving the speakers 11 feet further away I suppose that shouldn't be an issue in any case. If anyone tries the MOTU AVB interface with a current version of Mac OS I'd be curious to hear the results. I found info about problems with older El Capitan and Yosemite Macs but nothing about High Sierra or Mohave.

In my case I'd remote the MIDI over IP and use a fully functional console (no touch screens) if I needed to remote the console from the audio interface.

Re: MOTU 16a Thunderbolt vs AVB to Mac

PostPosted: Sun Dec 16, 2018 9:49 pm
by rhedgebeth
Yes, it is a PCIe card which will deliver sub millisecond latency. Were it not that the particular situation involves professional musicians in an academic situation we might otherwise be satisfied with the virtual sound card. As you accurately point out the difference amounts to being - at most - 11 feet farther from the speakers. This would not be very noticeable (if at all) to typical church musicians. There are other things more noticeable that get far less attention.