(17-May-2019, 21:50)alaw Wrote: [ -> ] (17-May-2019, 21:12)daniel.avasilichioaei Wrote: [ -> ]So, I am wondering: why using Roon RAAT, and not Roon AIR?
I am also wondering why Roon Core playing the same track is sending to Devialet almost a double amount of information for Roon READY versus Roon AIR?
I'm wondering the same, lol. I've gone back to AIR for some listening tests, works great.
I will leave aside the question of perceived quality difference since:
1) I don’t perceive any
2) I don’t know of any physical mechanism providing an explanation
This being said, there is one major difference between RAAT and Air: RAAT is designed to support multiple zones synchronization.
This alone is enough to explain the difference is throughput, especially in the Devialet to Roon direction.
In the case of Air, the only needed feedback is for flow control. I am not surprised that this is much less information and frequent than what is required by RAAT to achieve their synchronization.
So bottom line, unless you use multiple synchronized zones, there is very little advantage of RAAT over Air. The only one I can think of is there is more details on the audio path representation with SAM, subsonic filters etc appearing with RAAT and not with Air.
The drawback is the increased load on the network and for those unfortunate to experience the ‘slow loading issue’ drop outs and stuttering.
Jean-Marie
I have my Roon running on a MacBook Air which is at least five years old. However, I did buy it with an SSD and it had run quite satisfactorily until recently. I admit to never shutting down the MacBook unless compelled and find that quite often now when I load Roon it plays tracks for a few seconds then skips to the next. The solution so far is to reboot the machine and everything runs satisfactorily after that. No idea why, although memory corruption is the probable cause.
The 6.2Mb/sec is 0.62% of utilization of 1000Mb network. Why to worry about it?
(18-May-2019, 13:17)maxijazz Wrote: [ -> ]The 6.2Mb/sec is 0.62% of utilization of 1000Mb network. Why to worry about it?
I'm not worry about the bandwidth (quantity), but about the reason for 100% more traffic (quality).
I understand that RAAT is a much more laden network protocol than AIR, but I'm totally lost as to why 100mbps network speed (or wifi) would work but not gigabit speed. Hopefully, someone at Devialet and/or Roon does understand what is going on.
I believe AIR is still UDP while RAAT is TCP. TCP has much higher payload than UDP so for the same track playing one would observe higher bandwidth use with TCP. It's been a while since I looked at the difference in bandwidth use between the two, but I think TCP showed more than 50% higher bandwidth use than UDP. 100mbps is enough by a wide margin by the way - 192kHz material is less than 15mbps using RAAT (TCP). 1000BaseT and 100BaseT are quite different as I think the two are running different protocol stacks within the same hardware. The obvious difference is four pair use for gigabit vs two pairs for 100BaseT, but there are other differences too even if 1000BaseT is based on the 100BaseT spec.
(17-May-2019, 21:12)daniel.avasilichioaei Wrote: [ -> ]@David A: Excellent explanation. Thank you!
On my configuration via router, Roon READY was not working fine (mainly on HRs and MQAs). Roon AIR (bit-perfect asynchronous mode) had no problem at all.
Today I had setup Direct Ethernet (Gigabit) connection between my Roon Core (Windows 10 PC) and Devialet Expert, using a second network card. Then I spent many hours using/testing both Roon READY and Roon AIR... 
At the end, I came to two conclusions:
- Switching to Direct Ethernet (Gigabit) connection made an obvious improvement to Roon READY. Still not perfect, but it is almost acceptable.
- A possible cause for Roon READY issues may be the fact that this protocol is much "heavy" for the network (and devices) than Roon AIR. Since now I have a dedicated network card for Roon Core - Devialet communication, it is easy to compare. Thus (see attached pictures):
- Average network traffic from Roon Core to Devialet is almost double for Roon READY versus Roon AIR.
- Network traffic from Roon Core to Devialet is very fluctuating for Roon READY, while for Roon AIR it is pretty steady.
- There is ~8 times more network traffic from Devialet to Roon Core for Roon READY versus Roon AIR.
On the other hand, I didn't notice sound quality differences between Roon READY and Roon AIR.
So, I am wondering: why using Roon RAAT, and not Roon AIR?
I am also wondering why Roon Core playing the same track is sending to Devialet almost a double amount of information for Roon READY versus Roon AIR?
(18-May-2019, 16:24)daniel.avasilichioaei Wrote: [ -> ] (18-May-2019, 13:17)maxijazz Wrote: [ -> ]The 6.2Mb/sec is 0.62% of utilization of 1000Mb network. Why to worry about it?
I'm not worry about the bandwidth (quantity), but about the reason for 100% more traffic (quality).
Daniel,
I think some answers are now becoming apparent:
- the higher Roon to Devialet traffic is in large part probably due to the higher bandwidth requirements of TCP which RAAT uses and AIR doesn't.
- the much higher Devialet to Roon traffic is harder to pin down. It's clear the Devialet can be sending more information back to Roon with RAAT (tone control and SAM settings when they are in use, perhaps other info) and if TCP has higher bandwidth requirements then those higher requirements will be operating in both directions so that's part of it also but that doesn't explain why Devialet to Roon with RAAT is roughly 8 times higher than Devialet to Room with AIR when traffic in the other direction is only 2 times higher with RAAT than with AIR.
One thing which intrigues me is the difference in the flow of the traffic shown in your graphs, the fairly even flow with AIR vs the more episodic flow with RAAT. @
Confused points our that Devialet say "
Asynchronous BitPerfect mode provides optimal audiophile-grade conditions: music is transferred from your computer or smartphone to the amplifier with zero alteration to audio samples. This mode of transfer occurs at the exact same rate as Expert Pro digital-analog conversion rather than the speed imposed by your computer. The amplifier therefore dictates computer behavior rather than the opposite, with AIR intelligence split evenly between the two devices." The interesting part in that is the last bit which says "with AIR intelligence split evenly between the 2 devices. I'm wondering whether that means that with AIR the computer side uses it's knowledge of the resolution of the file to "push" the data to the Devialet at a regular rate whereas RAAT may require the Devialet to "pull" the data from Roon with regular "send me more" messages. If the Devialet has to initiate more communication with Roon regarding data transmission with RAAT than it does with AIR, that's going to be another cause of higher Devialet to Roon traffic and not only the number of messages being sent but the amount of data in those messages and the higher bandwidth requirements of TCP are going to play a part as well.
The bottom line in all of that is that it seems we have to expect higher traffic flows in both directions with RAAT because TCP has higher bandwidth requirements and there may also be extra traffic from the Devialet to Roon because the Devialet needs to send more data to Roon when using RAAT. Even with that higher amount of traffic in both directions, however, the demand placed on the connection are well within the bandwidth capabilities of both 100 baseT and gigabit ethernet connections so the bandwidth alone shouldn't be a concern for quality purposes.
After reading all very helpful replays to my previous test, I decide to make an additional test: for my dedicated network card for Roon Core - Devialet communication, I forced connection speed to 100/1000 Mbps, and repeat the test for both Roon READY and Roon AIR, using the same track (MQA 192kHz). Result is here attached.
My observations:
- Roon AIR acting in a very similar way for both 100/1000 Mbps connections (same amount and fluctuations for network traffic).
- On 100 Mbps connection, both Roon READY and Roon AIR are acting in a very similar way, and both are working fine with no issues at all. Indeed, ROON Ready is generating a very little more network traffic, and it is a little more fluctuating. Since we are speaking about different protocols, I believe that both differences are absolutely normal.
- On 1000 Mbps connection, Roon REDAY is getting crazy: it is generating a huge amount of additional network traffic, and also very big fluctuations.
As far as I can see, both protocols are fine, but there is a handshake issue between Roon READY and Devialet Expert on 1000 Mbps network.
Roon AIR:
100 vs
1000
[
attachment=3451] [
attachment=3452]
Roon READY:
100 vs
1000
[
attachment=3453] [
attachment=3454]
I haven't received any suggestions about better ways of setting the manual IP addresses in the method I described above but I have done a bit of experimenting on my own. What I have found is that things sound better to my ears and that some things like Roon searching for an artist or album work faster if you set the Gateway and DNS server settings for both the Nucleus/Nucleus+ and for the Devialet to 0.0.0.0 which effectivley means there is no Gateway or DNS server address. If you'e trying my method for a direct manual connection you may wish to change those 2 settings in the ethernet settings in both Roon and your Devialet's network configurations.
I've also started a new thread,
https://devialetchat.com/Thread-How-to-m...Expert-Pro, about how to connect a Nucleus and a Devialet using this method with a revised set of instructions for the method.
@
daniel.avasilichioaei
Those graphs do seem to show that something different is going on with a gigabit speed connection in your network when you use RAAT. I have no idea why there's that big difference especially compared to AIR but I wonder whether it has something to do with the "slow loading" messages and dropouts that some but not all people get with a gigabit connection.
I don't get those problems and I can't run any software of the sort you're using to get those graphs on my Nucleus+ so I can't run a similar comparison in my system.