Hifi news review Bartok Apex

Thanks Anupc!

Couldn’t agree more. Personally I wouldn’t pay for a DSD 256 captured file if it was converted to DXD for editing. I’d just buy the DXD master. According to Native DSD, most of their DSD titles are transcoded from DXD masters provided by their partnering record labels. I’ve heard of a couple of boutique labels doing tape transfers to high rate DSD or straight to DSD with no editing. In these cases I just buy the 128 version - many of which sound lovely, but really amount to 1% of my entire library. My personal opinion is the value of high rate DSD is on the ‘capture’ side.

Given the choice of playing DSD 128 titles on my dCS vs DSD 256 on one of my other chip based DACs, I’ll stick with the dCS.

1 Like

Got it. I see where you’re coming from now. Here’s the thing though; DSD64 vs. DSD128 is quite different from DSD128 vs. DSD256 - Noise-shaping on DSD64 is right at the edge of the audible spectrum, which DSD128 pushes out considerably. Whereas the impact of DSD256 is far less significant compared to DSD128.

There was a simplified visual of this posted by Andreas Koch that might explain it better (within a piece he wrote for PF in 2015 - a little dated, but well worth a read);

DSD64 already captures the entire human audible spectrum, so a higher resolution isn’t necessary other than to offer improved/simpler reconstruction filter response. For any given DAC, all else being equal, for example, the source master, then any benefits of DSD256 over DSD128 is likely to be all but lost in the overall system chain.

If you might recall, MQA was actually “threatening” to change the very face of all digital streaming, especially when Warner Music, Sony (and a couple of others labels) appeared to jump on board.

Even today, in many parts of the world (where Qobuz hasn’t reached), Tidal/MQA is the only available “high resolution” streaming source. So, dCS was absolutely right to implement it as fast as possible. The same cannot be said of DSD256.

Obviously I’m not at all opposed to dCS supporting DSD256, but at what cost to other more substantial areas of improvement?

Purely hypothetical, but if dCS asked you to choose between R&D spend on a “3.0 Mapper algorithm” versus “DSD256 support”, which would you chose? :laughing:


In a heartbeat, for my selfishness if I can’t choose “both” :wink:, I’d choose the improved Mapper, because I don’t really worry about dCS’s “techno-competitive posture.” And I completely agree that the difference between 64 and 128 is far more important than the difference between 128 and 256 and anything higher.

For me, the feature is about being able to play back as close to the original recording as possible. That is admittedly a very, very small subset of recordings.


If dCS think they could build a better mapper, then that would get my vote too, however market demands often push for ever bigger numbers.

When building products, it’s easy to get stuck between ‘What customers want’ and ‘What the manufacturer thinks will make customers’ lives better’, but not be able to do both


I’m going to keep my responses short, as I believe @Anupc’s technical knowledge (greatly) surpasses mine.


Yes of course. This is the law of diminishing returns. But why does this matter? Many dCS users want DSD256+ to see/test/explore if it is >1% better. Virtually all competing high end DACs offer this functionality Now. dCS, which claims to be the DAC leader, is behind in this feature. This is a statement of fact. Many on this forum have raised it before, and finally an audio reviewer has now also raised/written it, per this thread.


Ok, but this was not your original position. Your original position was that dCS only implements technologies with demonstrable sonic improvements. Any engineer exposed to MQA’s confidentially agreement (required for implementation) would immediately know that MQA was a lossy format. dCS implements features it perceives the market wants to sell more units (at least that’s what a qualified CEO should do). At one point that was MQA, hence the adoption. The future of high end audio is high res recordings. It doesn’t have a fancy name like “Master Quality Authenticated.” It has a different name: DXD and DSD.

DSD256 may be 0.0001% of current recordings, but it is ~40-60% of leading high res sites like NativeDSD. Hopefully dCS can soon roll out this feature.

It matters because resources are not limitless, and R&D spend will usually go to where it makes the biggest bang-for-the-buck. DSD256 is not it, at least in my opinion (and apparently others when asked to make a choice :wink:).

In any case, we can debate this point till the cows come home, only dCS can say about actual customer demand and development efforts. So, keep at it :grin:

I’m not a fan of MQA for various (mostly commercial) reasons, but I’d have to disagree with the technical characterisation of MQA.

For anyone with access to only redbook streaming sources, Tidal/MQA is absolutely a sonic improvement (for majority of the transfers)!

Arguments against MQA’s because it’s “lossy” or “time-smearing” suggests not really understanding how MQA works. Is it better than full lossless 192kHz/384kHz, of course not. Is it better than redbook 44.1/16 where high-res masters were available? Absolutely!


I am pleased that @Anupc hs raised the point :

The requests for DSD256 give the impression that any upgrade shall be free ( like v.2 of Vivaldi etc. or of a Mosaic version). However many of the posts are written written by relatively new members who may not have been here when dCS posted about the costs to them of alterations free to customers or otherwise.

It is not a case of a senior manager giving the instruction " Jack, press the button that releases DSD256 before you go home today". No, the upgrade first has to be selected from a long list of possible things to do and be costed. From dCS’ postings some years ago the likely sum is going to be in six figures. Irrespective of any technical discussion as to merit that investment would surely be excessive for :

A costly exploration.

Of course none of dCS’ competitors offer DSD256 as an upgrade. It is only a feature on the latest model. There eventually will be replacements for Bartok, Rossini and Vivaldi. These may well include DSD256. dCS could follow this path asap in order to meet your request but would you be prepared to sell your current model and buy the replacement just for DSD256?

For the moment an upgrade cost to dCS or to their customers would seem likely to outweigh both demand and need. It may be be there when new generations are released but IMO it does seem excessive for an upgrade right now.


I am in the camp that well understands that offering a feature like DSD256 is significantly more complicated than merely pressing a few buttons. And that it would utilize finite resources. I suspect most of our members here are well aware of that. However, this statement is not entirely true, and while I don’t want to derail this thread from its purpose of discussing the Bartok Apex review, it’s worth mentioning that PS Audio’s Direct Stream DAC, also a FPGA architecture, didn’t start out with 256DSD. It got there through upgrades. And they were free. As for “latest model,” that may be true in a technical sense, but for many DAC makers, the list also includes “the latest model’s predecessor, and its predecessor, and the predecessor before that, etc.” MSB has had it on its current models, several of which have been on offer for years. This isn’t “the latest thing.” It is officially old hat.

I’m not sure why this distinction should matter very much, as our excellent dCS representatives have acknowledged that the current architecture has adequate horsepower for the playback. In fact, calling out that distinction strikes me as providing more support for the upgrade, not less, because features of this type are part of the selling point of FPGA architecture. “Most other makers can’t do this; we can!”

Customers are in fact asking for this. Some have said they wouldn’t put it first on their list; I would always vote for better SQ, but the explanations of why those two features keep being positioned as competitive seem fairly thin to me. We’ve now been talking about this for over two years. In that time, the DSD256+ library has continued to grow. The communities external to our community have grown a bit louder in their criticism of the feature’s absence. I would now put the question another way:

Since the ability to provide significant upgrades is a touted feature of the FPGA architecture, and that architecture is both a selling point and significant SQ aspect of the system, why not deliver on one of the upgrades that the customers want and of which the system is capable?

It may not be important to everyone (not all features are), but its potential (system upgrade) is part of the raison d’être of the product design. DCS put that in stark relief with this week’s release of Lina 1.1:

If we talk anymore about this, it should probably be over in that DSD256 Feature Request thread. This is definitely not about the Bartok Apex review.


Not to side-track this thread further… but as I recall, PS Audio had significant limitations in supporting DSD256, for example, they could only do it via their I2S port and not the Ethernet port? (To be fair, if I’m not mistaken, the limitation was on the ConversDigital board rather than on the main DS DAC?)

By the way, has anyone else tried playing a DSD256 track directly via Mosaic lately? Give it a shot. You’ll hear “music” but with seemingly bit missing (I seem to recall in the past it used to be silence, so maybe we’re half-way there? :rofl:)

1 Like

That’s absolutely correct. There was also a specific OS requirement [Solaris IIRC].

Will have to check that out. :wink:

1 Like