Introducing Olympus & Olympus I/O - A new perspective on modern music playback

For those who just started reading up on Olympus, Olympus I/O, and XDMI, please note that all information in this thread has been summarized in a single PDF document that can be downloaded from the Taiko Website.

https://taikoaudio.com/taiko-2020/taiko-audio-downloads

The document is frequently updated.

Scroll down to the 'XDMI, Olympus Music Server, Olympus I/O' section and click 'XDMI, Olympus, Olympus I/O Product Introduction & FAQ' to download the latest version.

Good morning WBF!​

We are introducing the culmination of close to 4 years of research and development. As a bona fide IT/tech nerd with a passion for music, I have always been intrigued by the potential of leveraging the most modern of technologies in order to create a better music playback experience. This, amongst others, led to the creation of our popular, perhaps even revolutionary, Extreme music server 5 years ago, which we have been steadily improving and updating with new technologies throughout its life cycle. Today I feel we can safely claim it's holding its ground against the onslaught of new server releases from other companies, and we are committed to keep improving it for years to come.

We are introducing a new server model called the Olympus. Hierarchically, it positions itself above the Extreme. It does provide quite a different music experience than the Extreme, or any other server I've heard, for that matter. Conventional audiophile descriptions such as sound staging, dynamics, color palette, etc, fall short to describe this difference. It does not sound digital or analog, I would be inclined to describe it as coming closer to the intended (or unintended) performance of the recording engineer.

Committed to keeping the Extreme as current as possible, we are introducing a second product called the Olympus I/O. This is an external upgrade to the Extreme containing a significant part of the Olympus technology, allowing it to come near, though not entirely at, Olympus performance levels. The Olympus I/O can even be added to the Olympus itself to elevate its performance even further, though not as dramatic an uplift as adding it to the Extreme. Consider it the proverbial "cherry on top".

 

Hi Emile @Taiko Audio = When reversing polarity is it better for SQ to do it at the amp rather than the speaker? Thanks!

I'm not Emile, but it doesn't matter in the least.

Steve Z

 

Thanks very much, Wil for revisiting this.

I've been letting my Olympus play silently for the most part with my amp off to get some hours on the XDMI analog card. However, I finally broke down and snuck a little listen yesterday, just to hear how things were going.

Three hours later, I can say they are going very, very well.

I already can say with confidence that the quick comparison I did between Olympus USB and Extreme USB was no contest and based on that brief trial I've already sold my USB cable without a second thought.

However, as good as Olympus USB might be, when I started listening to Olympus XDMI analog it was game over for any thoughts of using Olympus USB for anything other than file transfers, not that I have any internal storage in my Olympus.

Olympus XDMI analog is doing everything better than I ever heard from Extreme USB > dCS Vivaldi APEX + Vivaldi Clock + Cybershaft OP21 master clock. Less noise, more clarity, nuance, space, imaging, ambience, the "three Ts" -- texture, tonality and timbre, articulation. And yes, even dynamics. Basically, any parameter a person would care to name is better, sometimes shockingly so, via Olympus XDMI analog.

I don't find it hard to believe that you find better dynamics with your Aires Cerat DAC. I don't have any experience with AC DACs but I do have an AC Incito S preamp here on long term loan through the generosity of Vassil (@nenon) and I what I hear and what I gather from reading about AC products in general is that dynamics are one of AC's fortes. However, I would also take Emile up on the experiment of reversing phase just to rule that out as a confounding factor.

It could be that the money shot for me might be Olympus XDMI AES/EBU into Vivaldi APEX, an experiment I will be getting to soon. I haven't heard anything about dCS being willing to work with Taiko on XDMI, so I'm not holding my breath on that one.

In the meantime, I'd have to say based on what I'm hearing with Olympus XDMI analog that no one who has posted rave reviews so far is exaggerating in the least. I couldn't be happier.

Steve Z

As we have been saying, another serious wake up call to DAC manufacturers. Writing with a small group of audiophiles tonight, and if these findings continue, "we will all want an Olympus". Seems like the smart call for DAC manufacturers to pursue partnerships with Taiko on XDMI. Kudos to Lampizator and MSB already (my 360 upgrade en route back to US landed LA today, Olympus order placed December 30, 2023 = #84 in queue). And wow, to think this is just the beginning. Thanks for this report out @oldmustang. Very compelling.

 

Hi Emile @Taiko Audio = When reversing polarity is it better for SQ to do it at the amp rather than the speaker? Thanks!

I have an inverting preamp so I merely reverse the polarity coming out of the amp. Instructions stated you can also do it at the speaker. It doesn’t matter

 

Hmm…. So you’re thinking Internal Olympus Dac may have reversed polarity relative my Aries Cerat Ageto pre? I’ll give it a try.

I do already have normal polarity reversed because (as I vaguely understand) the Areis Cerat Ianus Triodfet components are single stage and so need to be reversed before the speakers to be in correct polarity to the speakers.

XDMI is “in phase”, but a lot of gear is not as indeed every gain stage reverses phase. Your sonic descriptions reminded of reversed phase, and now your system setup description seems to suggest this may indeed be the case

 

I haven't heard anything about dCS being willing to work with Taiko on XDMI, so I'm not holding my breath on that one.

We have not heard back so most likely no.

 

Hi Emile @Taiko Audio = When reversing polarity is it better for SQ to do it at the amp rather than the speaker? Thanks!

Emile may have a different view, but my experience is that it does not matter, either at the amp side or at the speaker side, but not both, in order to switch polarity.

I always do amplifier side but agree with @Moladiego, shouldn’t matter, technically..

 

If you’re asking me, I wouldn’t bother. These things can change, in which case you’ll have wasted your time.

(@Taiko Audio )

Based on Ray's tests and listening sessions , (@ray-dude ), the sampling rate significantly affects power consumption, which in turn impacts sound quality (when using the Taiko XDMI-DAC daughterboard).

If I understand correctly, he notes a threshold around PCM 24-bit/96kHz.

Beyond that limit, sound quality begins to degrade.

That's what led me to consider down-sampling some of my music files.

However, am I right in understanding that this limitation might be addressed in future updates?

Are you suggesting that upcoming changes could potentially remove that limit?

What kind of changes would that take? A new XDMI-DAC daughterboard?

Of course, if I can avoid wasting time downsampling, I wouldn't say no ))

Cheers,

Thomas

EDIT :

As Ray suggests above, downsampling can be done upstream, and in real time, via Roon.

But isn't that just moving the problem upstream?

More processor time means more power, more heat, and therefore more noise.

 

Any update on shipping progress? The status page hasn't been updated in 8 days. Not sure if the orders listed have been shipped yet.

 

Any update on shipping progress? The status page hasn't been updated in 8 days. Not sure if the orders listed have been shipped yet.

Refresh your browser.

 

(@Taiko Audio )

Based on Ray's tests and listening sessions , (@ray-dude ), the sampling rate significantly affects power consumption, which in turn impacts sound quality (when using the Taiko XDMI-DAC daughterboard).

If I understand correctly, he notes a threshold around PCM 24-bit/96kHz.

Beyond that limit, sound quality begins to degrade.

That's what led me to consider down-sampling some of my music files.

However, am I right in understanding that this limitation might be addressed in future updates?

Are you suggesting that upcoming changes could potentially remove that limit?

What kind of changes would that take? A new XDMI-DAC daughterboard?

Of course, if I can avoid wasting time downsampling, I wouldn't say no ))

Cheers,

Thomas

EDIT :

As Ray suggests above, downsampling can be done upstream, and in real time, via Roon.

But isn't that just moving the problem upstream?

More processor time means more power, more heat, and therefore more noise.

Hi @SwissTom ,

There’s quite a bit more to it than that.

Let me start with explaining why power consumption rises with increases in sample rate.

Saving power is a critical part of today’s high performance processors. The more power they can save, the higher the clock frequencies they can reach, for longer durations.

Higher sampling frequencies translate to higher data transmission rates which lowers the CPU’s ability to save power, decreasing performance and increasing temperatures, causing the CPU to start throttling.

Now understandably the thought may cross your mind how can one of the fastest CPUs you can buy today run into throttling while performing such a simple task as music playback.

That is caused by how we utilise the system. I probably mentioned before XDMI is 75% software, 25% hardware. The way we transfer / process music is extremely inefficient from a computing POV. It is however very favourable for sound quality, and I’m going to leave it at that…

Now let’s take a look at what happens if we increase sample rates from a DAC perspective. If you closely examine the datasheet of the Rohm DAC chip we use, a few things may catch your attention.

Noise figures are negatively impacted, THD+N drops from 115dB at 44.1KHz to 105dB at 768 KHz for example. BCLK frequency doubles when sampling rate doubles, so do several types of noise, like phase noise. Current draw (power consumption) on the digital supplies doubles.

Now we have another aspect of the Olympus system design coming into play, which is the battery power supply. The output noise of the BMS/BPS is lower than even the lowest noise regulator you can design.

This changes things a bit as where you would normally expect the regulator output to be lower noise then the power rails which power it, and although increasing the current load does increase noise, this relatively matters less.

But now we have regulators increasing noise, the noise is actually completely dominated BY the regulator noise, and now all of a sudden doubling load makes quite a large difference.

I hope this provides you with a different perspective, the other side of the coin if you will, on some detrimental effects higher sampling rates can have on performance. This affects every part of digital playback, source, transport, interface and DAC. This then needs to be offset by a potential benefit of that higher sample rate, if that’s actually really there is a discussion for some other time.

As, again, XDMI is 75% software, it will most definitely evolve there, there is a very long to-do list on that alone, naturally you can expect things to change, performance will very likely increase, so will functionality.

Then we have the daughter boards as well, the same thing applies there, after redesigning the source, and now the interface, perhaps we should take a fresh look at how DACs are designed at some point in the future.

 

@Taiko Audio Thanks for that information. Very educational.

F/U question - does everything you described apply for both the internal as well as an external DAC connected via XDMI? I assume most does but I don’t know what may not.

 

XDMI is “in phase”, but a lot of gear is not as indeed every gain stage reverses phase. Your sonic descriptions reminded of reversed phase, and now your system setup description seems to suggest this may indeed be the case

If I understand correctly, the Internal Dac and Xdmi outputs to (In Phase)? If so, then based on all my Aries components being single stage, by removing the Aries dac, I am out of phase at my speakers:

USB (in phase) > Aries Dac single stage (phase out) > Aries Pre single stage (phase in) > Aries Amps single stage (phase out) > speaker cables reversed from normal for (phase in).

But for

XDMI (in phase) > Aries Pre (phase out ) > Aries Amps > ( phase in) > speaker cables reversed = (phase out)

 

@Taiko Audio Thanks for that information. Very educational.

F/U question - does everything you described apply for both the internal as well as an external DAC connected via XDMI? I assume most does but I don’t know what may not.

What is difficult to quantify is what these individual aspects contribute in differing system setups. Customer feedback / experimentation will be a big help in establishing that.

My results do largely but not fully mirror Ray’s results for example. For me SQ is good till 4fs, though I do ultimately prefer lower rates. I’ve however always felt that to be the case so from my perspective this is not new, but for sure much more obvious now.

 

Stravos, Aries Cerat, has just cleared it up for me:

Basically, it’s best just to try both ways because single stage does not always mean inverted and sometimes speakers themselves are inverted.

 

Hi @SwissTom ,

There’s quite a bit more to it than that.

Let me start with explaining why power consumption rises with increases in sample rate.

Saving power is a critical part of today’s high performance processors. The more power they can save, the higher the clock frequencies they can reach, for longer durations.

Higher sampling frequencies translate to higher data transmission rates which lowers the CPU’s ability to save power, decreasing performance and increasing temperatures, causing the CPU to start throttling.

Now understandably the thought may cross your mind how can one of the fastest CPUs you can buy today run into throttling while performing such a simple task as music playback.

That is caused by how we utilise the system. I probably mentioned before XDMI is 75% software, 25% hardware. The way we transfer / process music is extremely inefficient from a computing POV. It is however very favourable for sound quality, and I’m going to leave it at that…

Now let’s take a look at what happens if we increase sample rates from a DAC perspective. If you closely examine the datasheet of the Rohm DAC chip we use, a few things may catch your attention.

Noise figures are negatively impacted, THD+N drops from 115dB at 44.1KHz to 105dB at 768 KHz for example. BCLK frequency doubles when sampling rate doubles, so do several types of noise, like phase noise. Current draw (power consumption) on the digital supplies doubles.

Now we have another aspect of the Olympus system design coming into play, which is the battery power supply. The output noise of the BMS/BPS is lower than even the lowest noise regulator you can design.

This changes things a bit as where you would normally expect the regulator output to be lower noise then the power rails which power it, and although increasing the current load does increase noise, this relatively matters less.

But now we have regulators increasing noise, the noise is actually completely dominated BY the regulator noise, and now all of a sudden doubling load makes quite a large difference.

I hope this provides you with a different perspective, the other side of the coin if you will, on some detrimental effects higher sampling rates can have on performance. This affects every part of digital playback, source, transport, interface and DAC. This then needs to be offset by a potential benefit of that higher sample rate, if that’s actually really there is a discussion for some other time.

As, again, XDMI is 75% software, it will most definitely evolve there, there is a very long to-do list on that alone, naturally you can expect things to change, performance will very likely increase, so will functionality.

Then we have the daughter boards as well, the same thing applies there, after redesigning the source, and now the interface, perhaps we should take a fresh look at how DACs are designed at some point in the future. In all honesty, and with all due respect to everyone who has been involved with designing them, they all look somewhat outdated from where we’re at today.

(@Taiko Audio )

Hi Emile,

Thank you for your detailed and thorough response

It’s really interesting to get a glimpse of what’s happening on the other side of the mirror.

It adds some perspective and helps in understanding the issues at stake more clearly.

I’m genuinely impressed that you took the time to explain all of this!

I doubt any other company in the high-end audio would make the effort to be so educational.

Understanding things, even if only partially, is both interesting and reassuring.

Thanks for that!

Cheers,

Thomas

 

Hi @SwissTom ,

There’s quite a bit more to it than that.

Let me start with explaining why power consumption rises with increases in sample rate.

Saving power is a critical part of today’s high performance processors. The more power they can save, the higher the clock frequencies they can reach, for longer durations.

Higher sampling frequencies translate to higher data transmission rates which lowers the CPU’s ability to save power, decreasing performance and increasing temperatures, causing the CPU to start throttling.

Now understandably the thought may cross your mind how can one of the fastest CPUs you can buy today run into throttling while performing such a simple task as music playback.

That is caused by how we utilise the system. I probably mentioned before XDMI is 75% software, 25% hardware. The way we transfer / process music is extremely inefficient from a computing POV. It is however very favourable for sound quality, and I’m going to leave it at that…

Now let’s take a look at what happens if we increase sample rates from a DAC perspective. If you closely examine the datasheet of the Rohm DAC chip we use, a few things may catch your attention.

Noise figures are negatively impacted, THD+N drops from 115dB at 44.1KHz to 105dB at 768 KHz for example. BCLK frequency doubles when sampling rate doubles, so do several types of noise, like phase noise. Current draw (power consumption) on the digital supplies doubles.

Now we have another aspect of the Olympus system design coming into play, which is the battery power supply. The output noise of the BMS/BPS is lower than even the lowest noise regulator you can design.

This changes things a bit as where you would normally expect the regulator output to be lower noise then the power rails which power it, and although increasing the current load does increase noise, this relatively matters less.

But now we have regulators increasing noise, the noise is actually completely dominated BY the regulator noise, and now all of a sudden doubling load makes quite a large difference.

I hope this provides you with a different perspective, the other side of the coin if you will, on some detrimental effects higher sampling rates can have on performance. This affects every part of digital playback, source, transport, interface and DAC. This then needs to be offset by a potential benefit of that higher sample rate, if that’s actually really there is a discussion for some other time.

As, again, XDMI is 75% software, it will most definitely evolve there, there is a very long to-do list on that alone, naturally you can expect things to change, performance will very likely increase, so will functionality.

Then we have the daughter boards as well, the same thing applies there, after redesigning the source, and now the interface, perhaps we should take a fresh look at how DACs are designed at some point in the future.

As many know, I am a huge fan of PGGB, and like @ray-dude was, I am currently using it to upsample my most treasured albums to DSD512. On the Extreme, the benefits of PGGB's superior reconstruction and noise shaping/modulation is a no-brainer with my DAC.

But I do see where this improved accuracy can be swamped by the effects Emile has described. With Olympus XDMI, we are now in a very different ultra-low noise operating point.

My Olympus should be arriving within weeks, so I will soon be able to validate these findings for myself. Emile's explanation does give me a basis to understand them. And perhaps over time with future improvements, we may still be able to have our cake (ultra-low-noise Olympus XDMI goodness) and eat it too (superior reconstruction accuracy with PGGB).

As for downsampling, I encourage people to experiment and report here. The more data points we have the better.

 

My results do largely but not fully mirror Ray’s results for example. For me SQ is good till 4fs, though I do ultimately prefer lower rates. I’ve however always felt that to be the case so from my perspective this is not new, but for sure much more obvious now.

The other lever here is bit rate (16 bit vs 24 bit music). I suspect if we could easily do non-multiple sample rates and bit rates, the ideal for me would be between 3fs and 4fs, and a bit or two less than 16 bit

Couple more system tweaks and optimizations by Emile and team and that optimization point is sure to move

(all this with the caveat that my personal preference is very very strong for phase/temporal accuracy...others will opt to a different tradeoff point based on their systems and personal sound quality preferences)

 

The other lever here is bit rate (16 bit vs 24 bit music). I suspect if we could easily do non-multiple sample rates and bit rates, the ideal for me would be between 3fs and 4fs, and a bit or two less than 16 bit

Couple more system tweaks and optimizations by Emile and team and that optimization point is sure to move

(all this with the caveat that my personal preference is very very strong for phase/temporal accuracy...others will opt to a different tradeoff point based on their systems and personal sound quality preferences)

What sampling frequency does 3fs correspond to?

 

It doesn't exist in practice

If it did exist, it would be ~135kHz (1fs is 44kHz/48kHz)

I should qualify that my above estimates were to give a sense of where in a big range the tradeoff point could be. There is no practical way to operate there, I was offering my guess as a qualitative estimate to my ear/room/etc.

As a practical matter, I am limiting to 2fs (96kHz) 24 bit in practice, just so I don't have to think about this stuff

 

It doesn't exist in practice

If it did exist, it would be ~135kHz (1fs is 44kHz/48kHz)

I should qualify that my above estimates were to give a sense of where in a big range the tradeoff point could be. There is no practical way to operate there, I was offering my guess as a qualitative estimate to my ear/room/etc.

As a practical matter, I am limiting to 2fs (96kHz) 24 bit in practice, just so I don't have to think about this stuff

I need to review the formula.