Look forward to reading reviews and hearing impressions. Someone here is bound to have one inbound soon. @all2ofme ?
Unlikely — I’m not taking one for the team on this one!
None of my testing with servers (up to some lovely Innuos kit, but no further) has provided any tastiness whatsoever. Zero change that I could tell. Not worse, not better.
Now that might be because:
- I didn’t want it to make a difference. (I like not spending, despite what my bank statements might suggest.)
- I couldn’t understand why it should make a difference (1s and 0s and all that).
- It didn’t make a difference. (I’ve followed guidance from dCS on using a to-spec, unshielded Ethernet cable, and added a dose of network isolation wisdom from people who know a whole lot more than I do — hello @Anupc )
My server testing days are over, I suspect. If I hear positive murmurings from someone who stands to gain nothing from new approaches then I might look into it again, but the relaxing simplicity of my current state of affairs really is rather lovely
I think I’m in a similar place Ben. Although enough people that I trust report differences, even big ones, between servers that I am perfectly prepared to believe that this could well be a case of “we haven’t figured out what to measure yet” which is the flip side of “it shouldn’t make a difference”.
I find it funny however that they describe it as providing “a quite different musical experience than the Extreme” which many people (including a friend) regard as pretty much the top of the tree. Conventional wisdom is that advances in the state of the art are generally quite incremental. Well, the truth will out eventually. And if this is indeed the new king of the castle it will no doubt soon be Mike Lavigne who only swapped his Extreme for a Wadax a year-or-so ago.
Anyway, I am putting the popcorn on, this one will be fun to follow…
Here’s the thing, if you’re pulling down streams via an Ethernet/IP interface, then “bit transparency” is mutually exclusive to “Server A sounds better than Server B”; meaning, two Servers can’t both be bit-transparent and yet one sounds better than the other when streaming over a network. It’s not physically possible (in the real world).
So, as long as packets are not being lost, a Server has absolutely no consequence to the networking stack and packet re-assembly that goes on at the Streaming DAC end. If a Server has a sonic consequence, e.g. “sounds better”, then it must be changing bits, i.e. it’s not bit transparent.
Which is why Taiko Audio is full-of-shit and one of the biggest snake-oils in Audio.
Enjoy your🍿 @struts001
When a previous product is called the “Extreme” then naming options for the next ones grow narrower.
Olympus sounds high.
Everest sounds high.
Diamond sounds good.
Diamond Everest Olympus sounds good.
Just don’t sign ME up
This is one reason I enjoy dCS’s approach to model names. Vivaldi isn’t intrinsically higher ranking (as a name) than Rossini or Bartok. We apply that as consumers after the fact.
According to Taiko, the Extreme is “100%”, and Olympus+I/O is “300%”
So, if Olympus+IO is the bestest sounding, clearly it must be totally bit transparent. In which case, the Extreme can’t be, and people have been paying €28,000 for that!?
Incredible. They’ve built a seemingly successful business on pure hot air - I’m pretty sure they’re breaking some EU laws.
Personally I’m leaving the door open on this one as I haven’t compared different servers in my own system. I bought my sonicTransporter on a leap of faith seven years ago and haven’t (yet) looked back. At some point I am going to try to get an Innuos, Antipodes, Grimm or some such on home evaluation but that is not currently top of my priority list.
And I don’t believe in refuting subjective impressions with theoretical arguments because apart from being fallacious on a logical level it assumes that the theory is complete and correct. We used to believe that atoms were fundamental particles until Rutherford and Chadwick discovered protons and neutrons, the latter less than 100 years ago. It took another three decades before Gell-Mann proposed the existence of quarks, and another three before this had been ratified by empirical observation.
So this might all be snake oil and bit perfection may indeed be the only thing that matters up to the point of DAC**. But because it just might be that we don’t know what to measure to explain the difference (assuming it is not imagined) I will let my ears decide - and until then just enjoy the popcorn.
Having heard the extreme a few times in different very high end systems, i can say that when switching between it and others that cost considerably less, i couldn’t tell anything was different.
After all it’s only a computer, using cheap standard parts, with a fancy power supply in a fancy box.
I did find this YouTube video done on it year’s ago very funny.
That’s hilarious. How fitting that the system he shows at 7:36 is using a full Vivaldi stack (although I would point out that the box he has labelled as “Dac” is in fact the transport).
Yes indeed, Chord electronics too may have a problem in surpassing the name of their amplifier range “Ultima”.
What “theoretical argument” do you mean?
Everything I mentioned above about bit transparency from Servers driving packet streams are fully measurable. It’s not theory. Any lay person with some basic audio tools can derive clear objective empirical evidence.
And you’re making the standard audiophile mistake of comparing legacy physical interface problems (with SPDIF/AES) with how streaming works over networks.
That was amusing. I don’t want to comment on whether or not David’s comments on Taiko’s claims may or may not be correct.
Viewing from afar ( I am not in this market) I have found the pricing inexplicable. However having now seen the list prices of the items utilised in this video and that the cost to Taiko of that aluminium case and other costs are not accounted for and bearing in mind the industry rule of thumb that the average increase of cost from factory gate to retail sale is 5:1 in Europe then , irrespective of performance, this unit is going to cost consumers a lot ( as it does).
One thing that did raise my hackles was David referring to RAID as backup. It does not offer full backup and is rather for system continuance should a drive fail. Essential for e.g. an online bank but IMO not essential for a domestic audio system ( assuming a proper backup strategy).
You are absolutely correct. RAID is redundancy but not backup (unless configured for speed only - then it isn’t even redundancy).
A backup plan is like a spare tire for the car. You dont need one until you need one and if you dont have one you are dead in the water.
Now that collectively most or all of our info is stored digitally somewhere a backup plan is essential. I still refuse to use cloud storage services but really dont have a specific reason except any system like that can be hacked and will be hacked sooner or later.
Having a few hard disc crashes in my day I can speak from experience. The other thing about backups is you need to make sure they are happening as expected.
It is very easy to “set it and forget it” and then find out months later that a glitch in the chain caused the backup procedure to stop.
Thank you Anup, I think I understand your point, but I don’t think I’m explaining mine very well.
A digitally encoded wave form has only two dimensions, frequency and amplitude and clocking is only a factor at two points, ADC and DAC. I get that. But that is only taking account of what is going on at the logical level, i.e. the symbolic representation of ones and zeroes. That is what I mean by “theoretical”. You assert that there can be no possible difference because there is none that is explained by your conceptual model (digital audio) and you believe that model to be complete.
Servers and DACs are however electrical devices that exhibit behaviour that can’t be measured or explained within the conceptual realm of that symbolic representation, i.e. with digital theory.
So I know what we can measure, and we measure things that we know are relevant according to our conceptual model. But just like my analogy with quarks, we don’t measure things we don’t know about yet. One example would be leakage currents, people didn’t think to measure them in packet networks because they lay outside the conceptual model of digital (and yes, I know about transformer decoupling).
I am not saying that there are sources of noise or distortion that can be passed along with a bit perfect stream in a packet based network, just that there might be. Whenever you connect electrical devices together they interact electrically, regardless of whether they “identify” as digital or analog.
I know am unlikely to convince you, if for no other reason than that I am far from convinced myself. But I am leaving the door open in order to avoid making the standard audiophile mistake of thinking my knowledge is complete.
wow I’m still laughing. Seriously - is this thing UL or CE certified? the whole concept aside for the moment… Judging from the pictures there is some questionable thermal engineering. If those CPUs are indeed a pair of Xeon Scalables, it is unlikely the PSU and cooling is adequate to run full throttle. By comparison, the image below is how one workstation manufacturer deals with cooling of Xeon scalable processors. These processors by the way are designed for high performance compute workloads, not serving audio.
Look at it this way; I’m sure you’ve experienced pulling down a file via a Web browser. Do you expect the file to be in error when received?
No, of course not, because you understand that the bits are transmitted from the Web Server in packets of data across thousands of miles of the Internet and received on your PC with integrity checks and error correction to ensure that the packets are re-assembled back into the file as an exact replica of the original thousands of miles away.
Do you even care how powerful or not that Server is in order to receive the file error free? No, of course not.
Well, guess what. Your UPnP/Roon Server sitting in your home sends files across your home network to the Rossini’s S800 Streaming board - an ARM-based Single-Board-Compute platform running Linux by the way - in EXACTLY the same way!
People seem to know that even a $50 Raspberry Pi can do the job of file transfers across the Internet perfectly error-free, but yet somehow when it comes to streaming Audio, all bets are suddenly off, and they need to refer to Quarks and Infinity stones.
Thank you Anup. You clearly think I am suffering some basic cognitive impairment and that may well be the case. It could also be the case that our conceptual model is complete and this entire product category is snake oil and all the people who claim they have heard differences are just suffering a massive dose of confirmation bias and wearing the emperor’s clothes. Wouldn’t be the first time in this great hobby of ours! Your arguments make sense (just as those who asserted that protons and neutrons were fundamental particles also did).
No, not at all. Sadly though, you appear to be not much different from many other Audiophiles who love to conflate legacy audio technology issues, with those of network based streaming.
By the way, may I remind you of your post from a few days ago;
(Your point about James’ “theoretical arguments” is amusing - you do realise dCS does actual R&D and test everything? It’s not “theoretical arguments”).
More importantly though, you should heed your own advise about empirical testing instead of invoking elementary-particle physics
I don’t know if I am provoking you Anup. If so I apologise because it is really not my intention. And I really don’t want to get into a hair-splitting contest as others here seem to love to do. But please stop speaking to me as if I were a 5-year-old.
Thank you for the reminder but I see nothing paradoxical in what I said. James presents a number of conceptual arguments about different PLL bandwidths and what sorts of jitter and drift they are and aren’t susceptible to. These may be, and probably are, based on a lot of empirical testing but he doesn’t go into any of that. So as presented this is a theoretical rationale about causal factors. I am interested in empirical observations about the effects (does it actually sound better?)
I am aware that dCS does actual R&D. I am also aware that they perform extensive empirical testing, both objective and subjective, precisely because they understand that our conceptual models may not be complete (we may not know all the relevant things to measure) and/or that while our measurement instruments may not have sufficient resolution to measure things that matter our ears do.
Unless you want to have the final word I suggest we leave it there.
Haha. I used to think I knew everything too. Don’t worry, it goes over!