It’s a full blown switch, using a switch controller, we‘re only using 2 ports on it though.
Taiko Audio SGM Extreme : the Crème de la Crème
- Thread starter CKKeung
- Start date
@Steve Williams Steve, I'm no expert either. Perhaps, this is one of those instances of being better off knowing less. 
I do understand the value from isolating the segmented sub-network, filtering, etc. and keeping it to an 'n' of 1.
I do understand the value from isolating the segmented sub-network, filtering, etc. and keeping it to an 'n' of 1.
Steve williams
Site Founder, Site Owner, Administrator
All I care about is that you plug it all in as recommended and you push play and everything works. I really could care less as to semantics and proper definitions. If Taiko has given all this information as update , I take it as gospel but that’s just me.
FYI guys- seems it is both a media converter and a switch, no need to disagree
Technically it’s a switch, functionally it’s a media converter.
It would be much more beneficial to power the switch from the BPS then the router. We‘re assuming the switch being placed close to the Extreme and the router at a distance, likely where your internet enters your home.
It will be another month or so before we will disclose anymore details on that, our plate is full right now
Technically speaking a media converter should be working at Layer 1, a switch should be working at Layer 2, and a router should be working at Layer 3.
However, that hasn’t been the case for many years. Managed switches have Layer 3 capabilities. Routers are getting smarter and smarter and in some cases using technologies found on next generation firewalls that could work all the way up to Layer 7. Nowadays, it’s difficult to find a straight border line between all these network devices as they have a significant functional overlap.
There is no point to have a Layer 2 switch with two ports only. It’s just not practical, and I haven’t seen one yet. However, Emile has found a use case for such a device. It is using a switch chip, and so it should be classified as some type of a switch.
A “Media converter” has one simple function - to convert the signal from one physical media to another. The Taiko switch does not do that. Yes, it can convert from one media to another but that’s a function of the SFP transceivers, not the Taiko device.
The Taiko switch does some other things, though. As mentioned previously by Emile and captured in the PDF Christiaan uploaded earlier today:
“The Switch / network card combo reduces noise by offloading (moving) processing from the Extreme music server to the Switch.”
A media convertor cannot do that by definition. So, while it has only two ports, the switch chip must be doing something at Layer 2 or higher.
A two port switch is quite unusual implementation of a switch, and probably deserves a network device category of its own, but since one does not exist, I guess “switch” is the closest available category we can pick from. It's certainly much more than a media converter, and so I would not call it a media converter.
I don't quite understand why all that technicality matters. I suggest we put those disputes to rest.
It's an unique device and one of a kind. What matters is what it does to the sound. I placed my order today and looking forward to discover that. I have a bunch of other network solutions I will be comparing this to. It will be very interesting to explore that area again.
BTW, if you don't know what Layers I am talking about, I bet you also don't care if this device is called a switch, a media converter, or something else.
However, that hasn’t been the case for many years. Managed switches have Layer 3 capabilities. Routers are getting smarter and smarter and in some cases using technologies found on next generation firewalls that could work all the way up to Layer 7. Nowadays, it’s difficult to find a straight border line between all these network devices as they have a significant functional overlap.
There is no point to have a Layer 2 switch with two ports only. It’s just not practical, and I haven’t seen one yet. However, Emile has found a use case for such a device. It is using a switch chip, and so it should be classified as some type of a switch.
A “Media converter” has one simple function - to convert the signal from one physical media to another. The Taiko switch does not do that. Yes, it can convert from one media to another but that’s a function of the SFP transceivers, not the Taiko device.
The Taiko switch does some other things, though. As mentioned previously by Emile and captured in the PDF Christiaan uploaded earlier today:
“The Switch / network card combo reduces noise by offloading (moving) processing from the Extreme music server to the Switch.”
A media convertor cannot do that by definition. So, while it has only two ports, the switch chip must be doing something at Layer 2 or higher.
A two port switch is quite unusual implementation of a switch, and probably deserves a network device category of its own, but since one does not exist, I guess “switch” is the closest available category we can pick from. It's certainly much more than a media converter, and so I would not call it a media converter.
I don't quite understand why all that technicality matters. I suggest we put those disputes to rest.
It's an unique device and one of a kind. What matters is what it does to the sound. I placed my order today and looking forward to discover that. I have a bunch of other network solutions I will be comparing this to. It will be very interesting to explore that area again.
BTW, if you don't know what Layers I am talking about, I bet you also don't care if this device is called a switch, a media converter, or something else
.
Last edited:
For MSB it can be „very easy“ to implement. There are several parts in play here.
- The controller, for USB the vast majority uses an XMOS controller, this controller is USB in and I2S out.
- The transport between devices, this can be:
1) using a USB cable where the XMOS controller is located inside the DAC, like Lampizator and most other DACs really, the I2S signal is then generated inside the DAC.
2) using a separate external box in which the XMOS controller is located with an USB input and AES output (Totaldac reclocker) , or an USB input and I2S output, like the MSB Pro/ISL module, though that‘s a proprietary implementation of an I2S transport, if you’re interested in the specifics read up on the MSB website.
3) placing the XMOS controller inside the server, convert USB to I2S internally, and then transport over a different link to the DAC, like Wadax for example.
What we‘re doing is replacing that XMOS controller to rid ourselves of USB entirely.
For MSB we could provide direct PRO ISL out from the server, that is the easiest way.
For other DACS we could provide direct I2S out but transporting I2S over a distance is very problematic. That is why MSB went through the trouble of designing PRO ISL in the first place. So it would be better to take it further and add a new interconnect design, which we‘re aiming to have ready in May. This would require placing a small module inside the DAC, in a Lampizator DAC it could replace the JLsounds USB module for example and it would not need a high level of expertise to mount, we could probably find local people to do that about everywhere..
Yes we actually most likely will. In this case we‘d be looking for a degree market adaptation and this has been a very expansive project meaning it would probably end up retailing at the same 10-20K if only purchased by say just 20% of Extreme owners currently owning a DAC in that price range…
…”For MSB we could provide direct PRO ISL out from the server, that is the easiest way.“…
I hope your partnership achieve this goal. Indeed, the modularity of MSB architecture helps a lot.
Hi Emile,
what about the clock? Via USB the signal is clocked in the DAC close to the DAC chip. Will you send a synchronous signal from the Extreme to the DAC and/or will there be some feedback from the DAC clock to the Extreme like the ZEROLINK project of Soulnote and Sforzato in Japan?
Thx
Matt
Last edited:
I‘d go for a link not sensitive to clocking, atleast not in hardware, but that’s about everything I‘d share about it at this stage.
Emile
Is there a digital interface (I2S?) that is essentially Immune to Jitter/reclocking artifacts yet? Or in the near future?
In my experience, the fairly heroic MUTEC 'stack' that I still employ:
(2 cascaded MUTEC MC3+ USBs, each fed master clock signals from a MUTEC REF10 120SE).
part 1: nah@Steve Williams Steve, I'm no expert either. Perhaps, this is one of those instances of being better off knowing less.
I do understand the value from isolating the segmented sub-network, filtering, etc. and keeping it to an 'n' of 1.
part 2: i don't think a switch isolates the sub-network and that will come with the router.
in any case (cheers are appropriate here from those of you tired of me obsessing on this), i understand now that this started life as a full blown switch, offloads some functions that were previously done in the extreme (probably stripping off overhead that would be removed anyway in the extreme) and quiets things by restricting it to 1x1. i (finally) understand why it was called a switch, even if it no longer "switches"
[insert standing ovation that we can stop discussing this, at least from my perspective]
we now return you to your regular programming
