Taiko Audio SGM Extreme : the Crème de la Crème
- Thread starter CKKeung
- Start date
Very interesting thread. I just read all 48 pages...
The Extreme looks like an exceptional server. A quick glimpse of the pictures under the hood convinced me of the attention of every detail that has been put into building this device - good power supply, the EMI/RFI isolation for components inside the chassis, the EMI/RFI isolation between the chassis and the outside world, vibration isolation of the components inside chassis, the server motherboard with high speed / high core Xeon CPUs, the custom batch of high quality registered memory, the custom cooling, the chassis, the proprietary oscillator technology, the internal PCIe storage, Optane for the OS, etc. etc. The list goes on and on. It's really good to see a product like this.
As I was reading through 48 pages of posts, I kept thinking "does it get better than that?", "is there anything than can be improved?". Here are two thoughts that ran across my mind.
1. I have discovered that loading a track completely to RAM has its own sound quality benefits. That's why I stopped using Roon sometime ago. The PCIe storage in the Extreme is probably the best thing one can do for Roon playback, but I still think it could get a little better if loading to RAM first. I am curious if Taiko has experimented with that. No doubts they have paid attention to every detail here. Euphony / Stylus for me sounds better than Roon, but I am not using an Extreme server. It not only loads the track to RAM before playing but also copies the file to my Optane card first if the track is stored on my NAS. That makes NAS storage comparable to PCIe/Optane storage.
2. The minimalist in me, makes me think that USB is not the best interface for transferring digital from your computer to your DAC. There are too many conversions, clocking, etc. It can sound really good if done with care, but I like simple. In a way AES/EBU is more simple, but it also has its limitations. Many DACs would end up converting the SPDIF/AES/EBU signal to I2S at the end, so I think a good clean I2S signal out of the computer might be a better option for most DACs. If clean, clocked well, and transmitted cleanly (this is where LVDS helps), I2S signal could be superior to USB and SPDIF. YMMV and it is DAC dependant of course... I really wish server manufacturers and DAC manufacturers get together and create a unified standard and stop using USB in high end audio. Take the signal out of a PCIe slot and convert it to the unified standard that could be reclocked before the DAC or fed directly into the DAC without reclocking. I2S seems to be doing many things right, but it was designed 33 years ago and could use some refresh. I really like the MSB idea - proprietary I2S over fiber. And I really wish to see something like that going out of every music server into the DAC.
Imagine the case with MSB for example - your computer generates a USB signal that goes into the MSB Pro USB to be converted to their proprietary I2S over fiber... why not just generate the I2S fiber straight out of the PCIe on your computer?
I am very curious what our industry expert @Taiko Audio experience is on those two topics. Was the storage to Optane and then loading to RAM considered? Or was the design objective to make the best sounding Roon server? And was I2S output considered?
The Extreme looks like an exceptional server. A quick glimpse of the pictures under the hood convinced me of the attention of every detail that has been put into building this device - good power supply, the EMI/RFI isolation for components inside the chassis, the EMI/RFI isolation between the chassis and the outside world, vibration isolation of the components inside chassis, the server motherboard with high speed / high core Xeon CPUs, the custom batch of high quality registered memory, the custom cooling, the chassis, the proprietary oscillator technology, the internal PCIe storage, Optane for the OS, etc. etc. The list goes on and on. It's really good to see a product like this.
As I was reading through 48 pages of posts, I kept thinking "does it get better than that?", "is there anything than can be improved?". Here are two thoughts that ran across my mind.
1. I have discovered that loading a track completely to RAM has its own sound quality benefits. That's why I stopped using Roon sometime ago. The PCIe storage in the Extreme is probably the best thing one can do for Roon playback, but I still think it could get a little better if loading to RAM first. I am curious if Taiko has experimented with that. No doubts they have paid attention to every detail here. Euphony / Stylus for me sounds better than Roon, but I am not using an Extreme server. It not only loads the track to RAM before playing but also copies the file to my Optane card first if the track is stored on my NAS. That makes NAS storage comparable to PCIe/Optane storage.
2. The minimalist in me, makes me think that USB is not the best interface for transferring digital from your computer to your DAC. There are too many conversions, clocking, etc. It can sound really good if done with care, but I like simple. In a way AES/EBU is more simple, but it also has its limitations. Many DACs would end up converting the SPDIF/AES/EBU signal to I2S at the end, so I think a good clean I2S signal out of the computer might be a better option for most DACs. If clean, clocked well, and transmitted cleanly (this is where LVDS helps), I2S signal could be superior to USB and SPDIF. YMMV and it is DAC dependant of course... I really wish server manufacturers and DAC manufacturers get together and create a unified standard and stop using USB in high end audio. Take the signal out of a PCIe slot and convert it to the unified standard that could be reclocked before the DAC or fed directly into the DAC without reclocking. I2S seems to be doing many things right, but it was designed 33 years ago and could use some refresh. I really like the MSB idea - proprietary I2S over fiber. And I really wish to see something like that going out of every music server into the DAC.
Imagine the case with MSB for example - your computer generates a USB signal that goes into the MSB Pro USB to be converted to their proprietary I2S over fiber... why not just generate the I2S fiber straight out of the PCIe on your computer?
I am very curious what our industry expert @Taiko Audio experience is on those two topics. Was the storage to Optane and then loading to RAM considered? Or was the design objective to make the best sounding Roon server? And was I2S output considered?
Last edited:
Let me be the first to welcome you to the forum Vassils, and thank you for your kind words.
Yes we have experimented with playback from RAM, or even loading the entire operating system into RAM. It is a complex interaction of both positives and negatives. In our servers it nets out negative. Not just with the Extreme, but also in our previous generations of servers. We have put in a lot of effort on optimising memory utilisation, timing, its associated current draw behaviour and noise spectrum. Therefor it would not compare 1 on 1 with the results you have experienced.
The choice to design a dual CPU system was for a large part fuelled by finding a way around the sound quality impact Roons luxury interface has on sound quality. It does enable Roon processing to become virtually inaudible.
The choice of interface comes down to where you'd want your sensitive I2S signal to be generated. It is our opinion the best location for this is inside the DAC. The MSB Pro USB is an innovative and unique solution, halfway Server and DAC. I am a fan and the results cannot be denied.
Yes we have experimented with playback from RAM, or even loading the entire operating system into RAM. It is a complex interaction of both positives and negatives. In our servers it nets out negative. Not just with the Extreme, but also in our previous generations of servers. We have put in a lot of effort on optimising memory utilisation, timing, its associated current draw behaviour and noise spectrum. Therefor it would not compare 1 on 1 with the results you have experienced.
The choice to design a dual CPU system was for a large part fuelled by finding a way around the sound quality impact Roons luxury interface has on sound quality. It does enable Roon processing to become virtually inaudible.
The choice of interface comes down to where you'd want your sensitive I2S signal to be generated. It is our opinion the best location for this is inside the DAC. The MSB Pro USB is an innovative and unique solution, halfway Server and DAC. I am a fan and the results cannot be denied.
Thank you.
The music server is a complex system of hardware and software. If you improve one thing you most likely break something else. The overall end result is what counts at the end. I have no doubts, after reading all the stuff you have done, that you have found the best balance for your product.
It still bothers me that DAC and music servers manufacturers don't work together, though. One says - the best way for me is to get the sound out of the USB. The other says - I have to support USB to be compatible with everyone else. Then the new guys come in and say - well, everyone uses USB, I must do do same. And there you go - the USB became the standard just like that.
You have discovered that the PCIe storage sounds best. No surprise there. I would really like to see DACs fed by the PCIe slot on the computer, with as little conversion as possible, with external power of course and possibly over fiber optical connection, unless the DAC is super close.
But I am going offtopic here. Congratulations on the the excellent product. It looks amazing.
The music server is a complex system of hardware and software. If you improve one thing you most likely break something else. The overall end result is what counts at the end. I have no doubts, after reading all the stuff you have done, that you have found the best balance for your product.
It still bothers me that DAC and music servers manufacturers don't work together, though. One says - the best way for me is to get the sound out of the USB. The other says - I have to support USB to be compatible with everyone else. Then the new guys come in and say - well, everyone uses USB, I must do do same. And there you go - the USB became the standard just like that.
You have discovered that the PCIe storage sounds best. No surprise there. I would really like to see DACs fed by the PCIe slot on the computer, with as little conversion as possible, with external power of course and possibly over fiber optical connection, unless the DAC is super close.
But I am going offtopic here. Congratulations on the the excellent product. It looks amazing.
not fair to generalize about dac makers.
MSB has been 100% committed to optimizing every possible interface by (1) having their product be modular and so easy to change interfaces, and (2) to push the envelope on all the interfaces too. their Ethernet Renderer v2 was very good, their dual I2s for their transport is superb, and now the Pro USB is amazing.
i'm sure a few other dac makers are doing some similar things......
obviously i feel at least one server builder is pushing the digital interface window in a similar way.
Perhaps you missed my previous message, where I gave kudos to MSB for their initiative. I love the idea. Now if they open up the technology they are using to everyone and is good enough to become the new standard interface between DACs and computers, my dream would become true! I don't know what their intentions are but have a feeling they would prefer to keep it proprietary. I hope I am wrong, though.
And that other manufacturer you are mentioning is the only one that makes PCIe I2S cards. That's really good. And they even offer two levels of OCXO clock upgrades. The problem is that the CM8888 chip they are using is extinct and those cards don't support DSD.
It's probably not the right place to discuss, but I find the computer to DAC interface a big problem. The only reason I even mentioned this here, was because given the amount of engineering I have seen by Taiko in this thread, I am sure they have thought about that problem and may have some interesting comments. But equally, given how they have tackled every big problem on their way (i.e. Roon sound quality, noisy SSD storage, transformer vibrations, etc.) I have a feeling they just decided to go with the industry standard, add the two most common interfaces, and give the best of them to make those interfaces sound better than ever. What else can they do? Try to pioneer new standards and fight the industry
?
And that other manufacturer you are mentioning is the only one that makes PCIe I2S cards. That's really good. And they even offer two levels of OCXO clock upgrades. The problem is that the CM8888 chip they are using is extinct and those cards don't support DSD.
It's probably not the right place to discuss, but I find the computer to DAC interface a big problem. The only reason I even mentioned this here, was because given the amount of engineering I have seen by Taiko in this thread, I am sure they have thought about that problem and may have some interesting comments. But equally, given how they have tackled every big problem on their way (i.e. Roon sound quality, noisy SSD storage, transformer vibrations, etc.) I have a feeling they just decided to go with the industry standard, add the two most common interfaces, and give the best of them to make those interfaces sound better than ever. What else can they do? Try to pioneer new standards and fight the industry
?[QUOTE="Mike Lavigne, post: 587837, member: 57"(...) MSB has been 100% committed to optimizing every possible interface by (1) having their product be modular and so easy to change interfaces, and (2) to push the envelope on all the interfaces too. their Ethernet Renderer v2 was very good, their dual I2s for their transport is superb, and now the Pro USB is amazing.
(...) [/QUOTE]
Mike,
According to digital rules, if one interface is very good, the other is superb and another is amazing they are not optimized. Optimized digital interfaces should sound the same, digital has different rules than analog ...
(...) [/QUOTE]
Mike,
According to digital rules, if one interface is very good, the other is superb and another is amazing they are not optimized. Optimized digital interfaces should sound the same, digital has different rules than analog ...
Mike,
According to digital rules, if one interface is very good, the other is superb and another is amazing they are not optimized. Optimized digital interfaces should sound the same, digital has different rules than analog ...
the concept that every digital interface should sound the same is just silly. then dacs and servers should all sound the same too.
bits are just bits?
no, not really.
different approaches and degrees of execution and likely infinite variances in combinations of variables, as well as the human factor in perception will insure nothing is ever exactly the same. will we always hear those differences? (no) and do they matter (many times yes)?
Where are these rules of which you speak Micro.Mike,
According to digital rules, if one interface is very good, the other is superb and another is amazing they are not optimized. Optimized digital interfaces should sound the same, digital has different rules than analog ...
Good digital systems prove every bit as responsive to change as the good analogue ones. The higher you go up the resolution tree in a system the more obvious even the small changes become. The more generic and less revealing systems in either the hardware or in the replay of software tend to be the more lacklustre. If everything you do just sounds the same then the system most probably sucks.
PS and I always thought it was more analogue rules... like forever
Last edited:
Where are these rules of which you speak Micro.
Good digital systems prove every bit as responsive to change as the good analogue ones. The higher you go up the resolution tree in a system the more obvious even the small changes become. The more generic and less revealing systems in either the hardware or in the replay of software tend to be the more lacklustre. If everything you do just sounds the same then the system most probably sucks.
PS and I always thought it was more analogue rules... like forever
Digital storage and transmission at the top quality should be " bits are bits". We can romantically disguise our technical ignorance about correlation between digital implementations of digital data transmission and subjective sound quality comparing to the analog case, but we are comparing the incomparable.
The bits for bits debate surely applies to DACs - when digital data is converted to an analog signal - but in theory should not apply to servers . Digital data can't depend on physical properties of storage if staying in the digital format.
Please note that this debate can not be centered just on the servers, it is mainly correlated with the digital receiver of the DAC. I would love to have to have the opinions of MSB people on these matters.
People often consider that more responsive systems have always higher sound quality - not true IMHO, YMMV.
It seems to me that the people who implement the best systems are constantly finding differences between various implementations that outcome in changes to sound throughout the digital pathway.
Not sure where the technical ignorance actually is but certainly none of this bits are bits business ever actually plays out in reality... more like noise is noise and the bits is bits is more just theory much like the early digital perfect forever notions. IME as well as IMO.
I enjoy and learn from the kinds of discussion driven here especially by the people like Emile that are actually working day in and day out at the cutting edge of this technology and it seems to me that the thing that keeps being reinforced is that most everything seems to matter particularly when people are talking about actual experiences rather than just theories Micro. Ideas themselves can be marvy but the best understanding tends to come through practice and praxis IMHO.
Last edited:
I'm not interested in enaging into a discussion on this, but I'm sure you are aware that these "bits" are all transferred as a varying voltage on copper transmission lines, susceptible to everything that distorts analogue signal line levels, even more so.
As for vassils I2S comments, I don't think it's a particularly good approach to generate an I2S signal, consisting of data, word clock and a bit clock line, and transfer all of that over long sensitive copper interconnects with a multitude of impedance mismatches along the way and a general strong sensitivity to transmission line quality. What I admire about the MSB Pro solution is they generate I2S external to the server, convert and transfer these over fiber, using 1 fiber for data and the other for clock signals, this is much more insensitive to disturbances then copper based interconnect systems. Let's call it I2S done right.
It's worth noting that some types of jitter are perceived as improving sound quality. Which certainly adds a twist to any technical superiority discussions. This can enable technically inferior solutions to be touted as sounding better.
Everyone should know that digital information is carried by analog lines . In theory every time a bit is stopped in time all his previous analog distortions should disappear, however when we send it again we go again through this process, surely creating new "distortions".
The Extreme surely does a great job managing to do it in audio real time with top sounding characteristics using current DACs. It would be nice to go in some details of the process, particularly in the individual DAC matching, but you are right, it seems audiophiles prefer to stay in the empirical reign.
Nice to see you also aim at "best understanding" . It is what leads my questions. But IMHO practice and praxis without a proper model and theory is not the way to go in this digital audio business. Remember that using just practice and praxis it took twenty years to move from "bits are bits" to "minimal jitter is all that matters" ...
Last edited:
Surely theory is an early, valuable and incomplete phase of process... the life cycle of understanding moves much further through to execution and reflection. There is no digestion just in theory, implementation can create understanding beyond any initial theory... and that’s not even the end there. After that there’s still the letting go of knowing and the final dissolution to come. Bits are never whole. Understanding always is.
Last edited:
Here is the movie about TotalDAC [d1-Twelve MK II] & Taiko Audio [SGM Extreme].
This Movie was made by Korean Extreme customer Mr. Chun (Muscle Pain).
Although Korean explanation is too much, you can hear sound about d1-Twelve MK II & SGM Extreme.
As you know, in movie, there is MP 64-32, but you can't hear the sound.
Please ask why MP 64-32 is not playbacked~!!
Mr. Chun said SGM Extreme & d1-Twelve MK II sound is no. 1.
Never heard before~!!!!
Enjoy it~!!!!!
<iframe width="720" height="407" src="
" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>
This Movie was made by Korean Extreme customer Mr. Chun (Muscle Pain).
Although Korean explanation is too much, you can hear sound about d1-Twelve MK II & SGM Extreme.
As you know, in movie, there is MP 64-32, but you can't hear the sound.
Please ask why MP 64-32 is not playbacked~!!
Mr. Chun said SGM Extreme & d1-Twelve MK II sound is no. 1.
Never heard before~!!!!
Enjoy it~!!!!!
<iframe width="720" height="407" src="
Mike,
IMHO putting servers and DACs in the same pot in this discussion is silly. Digital transmission and storage should teoretically be able to sound the same, DACs surely not.
Surely we do not have all the answers. Perhaps you will find someday that your optical link improved the sound of your analog sources, and we will look for a reason for it.
Bits are not just bits all the time, but sometimes they are just that. I am just trying to debate where in the chain this transition happens.
Last edited:
Mike,
IMHO putting servers and DACs in the same pot in this discussion is silly. Digital transmission and storage should teoretically be able to sound the same, DACs surely not.
Surely we do not have all the answers. Perhaps you will find someday that your optical link improved the sound of your analog sources, and we will look for a reason for it.
Bits are not just bits all the time, but sometimes they are just that. I am just trying to debate where in the chain this transition happens.
honestly i'm more interested in how that AF1 Premium in your room compares to the EMT 927 you have (or had) than some digital theory <-> dac/server/interface conundrum of what is what with digital. i know it's a different thread......but sometimes you break some rules for the best reasons.
honestly i'm more interested in how that AF1 Premium in your room compares to the EMT 927 you have (or had) than some digital theory <-> dac/server/interface conundrum of what is what with digital. i know it's a different thread......but sometimes you break some rules for the best reasons.
The EMT 927 already moved to a new owner and it is still too soon to write about a night and day difference ...
BTW my question was inline with the thread debate, that surely must includes digital noise. Anyway it is interesting that most people here appreciate to debate all kind of theories on analog but seem uneasy with digital ...
I get the impression that folks on this thread don’t subscribe to your bits = bits “thing” nor have the appetite for such a debate.
P.S. Congrats on your new TT. Curious also to hear your listening impressions vs the EMT.
