Taiko Audio SGM Extreme : the Crème de la Crème

[Taiko Audio]

2 more datapoints:

1) In our previous place Fiber and Copper privided a different perspective, one not necessarily better then the other. In our current place, Copper is better, with a sizeable margin. Makes one wonder if “galvanic isolation” is a thing at all, especially considering the considerable increase in amount of devices and traffic we have now.

2) We have a dozen or so IP cameras all around the building recording HD video 24/7, about 5-6 MBs continuous traffic from those alone.

 

[Taiko Audio]

Thank you Emile. Amazing how many aspects play a key difference.
I will test busy network also. Will connect gaming server to my audio network.

I agree with 5 G wifi. It could be very nasty but if properly done you gain lightning speeds And sharper ,
better controlled sound.

Streaming HD movies, or say Netflix 4K would be a good test. Please share the results if you do.

 

[austinpop]

I know @nenon, @Taiko Audio, and some others here are aware of this, but I thought I'd just remind everyone of this.

As you ponder how busy/idle the network is, it's important to keep in mind that its impact on SQ is defined by what appears at the network interface(s) of the Extreme. Note that the Extreme will only see Ethernet frames of 3 categories:
- those that are specifically addressed to/from it (unicast)
- broadcast frames
- multicast frames.

Importantly, the Extreme will not "see" unrelated traffic between other devices in the network, as those frames will be filtered out by the switch(es)/router upstream of the Extreme. So while other busy devices in your premises should not directly affect the Extreme, they perhaps could do so for these reasons:
- if these busy devices result in an increase in broadcast or multicast traffic.
- if these busy devices result in an increase in "noise" in upstream switches/router.

All this said, I am very happy to continue practicing my "hit play and disconnect network" strategy.

 

[austinpop]

For those technically inclined, this book chapter explains things in a fairly simple and clear way: https://www.oreilly.com/library/view/ethernet-switches/9781449367299/ch01.html

The key sections that cover the filtering I talked about in my post above are titled "Address Learning" and "Traffic Filtering."

 

[Taiko Audio]

For those technically inclined, this book chapter explains things in a fairly simple and clear way: https://www.oreilly.com/library/view/ethernet-switches/9781449367299/ch01.html

The key sections that cover the filtering I talked about in my post above are titled "Address Learning" and "Traffic Filtering."

I’m very interested if anybody can make out a difference from deliberately loading their network / router / switches in a more continuous fashion, and if so, with negative or positive results, this experiment is more likely to be useful with copper connections.

 

[dminches]

I will make an effort to start streaming movies after my system is warming up and the sound has settled in. Maybe I will hear a difference.

 

[kswanson27]

I will make an effort to start streaming movies after my system is warming up and the sound has settled in. Maybe I will hear a difference.

I'll do the same. The trick will be to get my wife to stop watching Netflix/MHZ/Prime etc so I can compare with and without a busy network.

 

[cat6man]

I know @nenon, @Taiko Audio, and some others here are aware of this, but I thought I'd just remind everyone of this.

As you ponder how busy/idle the network is, it's important to keep in mind that its impact on SQ is defined by what appears at the network interface(s) of the Extreme. Note that the Extreme will only see Ethernet frames of 3 categories:
- those that are specifically addressed to/from it (unicast)
- broadcast frames
- multicast frames.

Importantly, the Extreme will not "see" unrelated traffic between other devices in the network, as those frames will be filtered out by the switch(es)/router upstream of the Extreme. So while other busy devices in your premises should not directly affect the Extreme, they perhaps could do so for these reasons:
- if these busy devices result in an increase in broadcast or multicast traffic.
- if these busy devices result in an increase in "noise" in upstream switches/router.

All this said, I am very happy to continue practicing my "hit play and disconnect network" strategy.

Rajiv
I can think of another possible mechanism here.
Network traffic, even if not seen at the extreme's input, can potentially impact SQ. If the other network traffic impacts the burstiness of the data into the extreme, it may impact sound quality, for example if it changes buffering processes. I think most will agree that buffering and flow control are important.

I also like the idea of music with and without Netflix streaming and will try that test as well.

 

[oldmustang]

I will make an effort to start streaming movies after my system is warming up and the sound has settled in. Maybe I will hear a difference.

I'll give it a try too, though streaming movies and downloading large files simultaneously pretty much brings my so-called high-speed DSL connection (a blazing 12 mbps down/2 mbps up) to its knees. Since my streaming/download speed is maxed out long before my local in-home network speeds are appreciably stressed, perhaps I should start a NAS-to-NAS backup on my network while doing the experiments.

Steve

 

[austinpop]

Rajiv
I can think of another possible mechanism here.
Network traffic, even if not seen at the extreme's input, can potentially impact SQ. If the other network traffic impacts the burstiness of the data into the extreme, it may impact sound quality, for example if it changes buffering processes. I think most will agree that buffering and flow control are important.

I also like the idea of music with and without Netflix streaming and will try that test as well.

Yes certainly, the load on the upstream switches, and particularly the router, can change the timing of the data flowing through the Extreme. By all means, do try the Netflix experiment.

I would just point out that this would be more of a concern with a "naïve" music server like Roon, which blithely transfers music bytes and chats on the network during music playback. I would think it would be much less of an issue with an intelligent music player like TAS/XD. One of the optimizations in TAS is to do all its network "business" before playback commences, pulling down the music files from the streaming files before playback. With the TAS/XD optimizations, the only network pollution would be the "ambient" or idle flow of packets to the Extreme. Even the TAS app is designed to not be chatty during playback.

So with that said, if there is indeed an SQ degradation with heavy activity upstream (the Netflix test), then I would be inclined to attribute it to increased physical noise in the upstream device, rather than burstiness. Oops, I started down the rabbit hole. I'm heading back. Bye.

 

[austinpop]

perhaps I should start a NAS-to-NAS backup on my network while doing the experiments.

That would actually be a better (more stressful) test!

 

[oldmustang]

That would actually be a better (more stressful) test!

My thought exactly, Rajiv. Thanks.

Steve Z

 

[cat6man]

Yes certainly, the load on the upstream switches, and particularly the router, can change the timing of the data flowing through the Extreme. By all means, do try the Netflix experiment.

I would just point out that this would be more of a concern with a "naïve" music server like Roon, which blithely transfers music bytes and chats on the network during music playback. I would think it would be much less of an issue with an intelligent music player like TAS/XD. One of the optimizations in TAS is to do all its network "business" before playback commences, pulling down the music files from the streaming files before playback. With the TAS/XD optimizations, the only network pollution would be the "ambient" or idle flow of packets to the Extreme. Even the TAS app is designed to not be chatty during playback.

So with that said, if there is indeed an SQ degradation with heavy activity upstream (the Netflix test), then I would be inclined to attribute it to increased physical noise in the upstream device, rather than burstiness. Oops, I started down the rabbit hole. I'm heading back. Bye.

Good observations, but if I were betting, a best in class flow control strategy is the big advantage that the taiko switch will provide, with taiko in control of both ends of the data flow into the extreme.

It's just an (educated?) guess on my part at this point. We are all learning here. It's hard not to go down the rabbit hole, but ill also try

Cheers (and stay safe)

 

[Taiko Audio]

I'll give it a try too, though streaming movies and downloading large files simultaneously pretty much brings my so-called high-speed DSL connection (a blazing 12 mbps down/2 mbps up) to its knees. Since my streaming/download speed is maxed out long before my local in-home network speeds are appreciably stressed, perhaps I should start a NAS-to-NAS backup on my network while doing the experiments.

Steve

NAS to NAS is interesting too. Managed switches can have support for more advanced features, like QoS, flow control, jumbo frames, spanning tree, these need to be configured on both the switch and the network adapter. This all has an "effect".

 

[Taiko Audio]

So with that said, if there is indeed an SQ degradation with heavy activity upstream (the Netflix test), then I would be inclined to attribute it to increased physical noise in the upstream device, rather than burstiness. Oops, I started down the rabbit hole. I'm heading back. Bye.

I guess we can have "burstiness" from a heavily loaded switch too, agree about that rabbit hole, it is hard to quantify the mechanics of all of this. Our router is most certainly not just sitting there turning thumbs:

 

[Muni]

just happy to share that I ordered an EXTREME and am happy that I did - though never heard one - ordered it because you all shared your experiences and joy for listening...
Thank you for that - and the waiting time will just fly away since the joy is taking over...
will let you know in about 8 weeks -

 

[Taiko Audio]

just happy to share that I ordered an EXTREME and am happy that I did - though never heard one - ordered it because you all shared your experiences and joy for listening...
Thank you for that - and the waiting time will just fly away since the joy is taking over...
will let you know in about 8 weeks -

Congratulations, and welcome to the forum!

 

[Muni]

thank you, Emile!

 

[Steve Williams]

just happy to share that I ordered an EXTREME and am happy that I did - though never heard one - ordered it because you all shared your experiences and joy for listening...
Thank you for that - and the waiting time will just fly away since the joy is taking over...
will let you know in about 8 weeks -

Just keep your seat belt on and your seat back in an upright position because Emile and his team are going to take you on the ride of your life

Welcome to WBF. This is my favorite thread on WBF. It goes with my morning coffee every day

 

[oldmustang]

NAS to NAS is interesting too. Managed switches can have support for more advanced features, like QoS, flow control, jumbo frames, spanning tree, these need to be configured on both the switch and the network adapter. This all has an "effect".

I guess that to some extent any conclusions we might reach about the audible effect of overall traffic, "burstiness", etc will be specific to our unique LAN configurations. In my case the NAS-to-NAS traffic I'd be using to load up the network comes "downstream" of an unmanaged switch and copper/fiber/copper conversion:

ISP > DSL modem/router > copper > OpticalModule > fiber > Extreme

Same modem/router > copper > FMC > fiber > FMC > copper > Netgear GS105 switch > all the rest of thehousehold and general computing stuff, including two NAS.

A different configuration might have a different degree of susceptibility to and audibility of network activity.

Steve Z