Life after Yggdrasil: Watering the Ash

Yep, because building an ASIC to do the very same thing is way more expensive than they can afford for such low-volume products. But the FPGA is doing nothing that an ASIC can't, beyond being reprogrammable. Hell, if you had enough I/O pins and a fast enough cycle time you could do it in software coupled with a bog-standard CoTs micro-controller.

 

He was able to do that because he had lots of unused capacity (cores/fabric) on the FPGA. No doubt he found some more efficient ways to employ them too (we almost always figure out a better way to do it for "next time" with any changes/development work).

Good on them for over-specifying their FPGA to allow for such updates though. Linn went the same way.

 

Cool, nice they don't scale price upward with moore's law like certain other companies -- ooooo new double capacity FPGA at 1/2 the cost, let's double the 500K taps with a line of code ;D -- good stuff, but wow they are pushing it with the price

 

It does seem to be their thing. Lots of overlap across many of their otherwise-complimentary products. That makes a bit more sense in their lower-tier products. But the flagships should, as far as I'm concerned, be more focused.

The Pro iCAN might be "feature rich" but fundamentally it's an amplifier.

The Pro iDSD is a DAC/streamer*/amp.

And if you don't want the streamer or, in particular, the amp (which is necessarily either compromised compared to the Pro iCAN or renders it completely redundant) you're paying for functionality that could easily be external. That pushes the comparison against other DACs into a higher price tier than would otherwise be the case.

Fundamentally it's the same issue I had with the dac1541 ... lovely DAC, but I don't need or want the headphone output or pre-amp functionality, and paying for shit I don't use has never sat well with me.

---

*And LinkPlay as the WiFi streamer-on-a-card implementation? That's the best they could do with all the time they've spent fannying about with the thing? A solution features in all manner of no-name products that, well, you've never heard of. Doesn't even support Roon. But, oh, golly, it does AirPlay and Spotify ... such crucial concerns for customers of $2,500 DACs. Particularly when that could be supported with a $100 external box AND get you Roon AND AirPlay 2 AND TIDAL support if you wanted it ...

 

I'm pretty sure Søren added those features to the dac1541 because they were virtually free because of it's architecture with the only added costs being the ports themselves.

 

And the board space, casing requirements (more complex panels cause disproportionate issues/costs), and R&D/QA time to implement and make sure those added features work correctly. "Virtually free" has a bad habit of turning $10 of parts, into $50 on the overall per-unit cost and a 5x multiplier on that for the retail price.

No idea what the actual numbers are in this case; when I raised the issue before "not much" was the answer. Tells me nothing useful. Regardless, I still don't like paying for things I won't use. Nor do I really want them on a product if I won't use them - at best they don't make things perform worse - but they're certainly not improving anything either.

But that's just my personal perspective.

 

I actually understand your point of view but feel the opposite. For instance, I don't understand why people would buy the Schiit Modi 2 when the Modi 2 Uber is such a better deal for only $50 more. Even if you don't need the functionality now, things can change and all that.

But that's just me and I can see where you're coming from.

 

They know they're never going to use the S/PDIF inputs?

There's a difference between choosing to buy a more-feature-rich version of an otherwise identical product and having the only option be to pay for functionality you know you'll never use.

I've learned, through experience, that buying things either "just in case" or on "futures" (promised features/updates) very rarely works out in my favor. May well be different for others.

And it might be different if there's a better than even chance that the additional functionality will get used in the relatively near future. Otherwise, I've generally found it preferable to wait to buy things until I need them. Too many things can change in the interim - especially where technology is concerned.

 

I tend to agree with what Torq is saying about unused features, but I also wouldn't passover a particular item because of it. I don't know what the headphone amp adds in terms of cost for the dam1541, but even with its inclusion it is around the same price as a Gungnir MB with comparable (arguably better) performance.

 

I happen to be in the sweet spot since I use the DAC1541 as a DAC, preamp for active monitors and headphone amp for both music and tv. The remote even lets me switch DAC sources. There is not a Schiit stack that does all that without duplicating components in redundant ways.

Am I the right target market for this unit? Is it the right strategy for Soekris? Fortunately for me I don't need to answer that. I just enjoy the 'fit'. If it doesn't fit your use case, well, you just get to choose something else.

 

This post may only be useful as a curious fact, but the Wadax Atlantis DAC uses an ASIC ($$$)

 

Definitely interesting!

At $69,000/unit they have a lot of leeway in terms of how many units the need to sell to cover the design, setup and fab costs of an ASIC. It's also worth bearing in mind that, depending on what's IN (or ON) any given ASIC, the first two costs, design and setup, can be minimized greatly.

For example, they could be using a slightly-modified existing design, which means it's principally fab-costs that come into play. In such cases, particular if they're on a process-stepping (larger feature size, say 28nm or so) that's simple photo-lithographic compatible, then their chips can be fabricated on the same line as the "base" chip for a much larger market.

But if it's a ground-up, totally custom, ASIC then that's a ballsy move and quite commendable. It would also indicate that the world-wide market for $69K+ DACs is quite a bit larger than I'd have wagered.

 

I wonder what sort of filtering they're running that requires an ASIC, unless they did it "because we can".

Feed-forward is about as specific as a "Turbo" badge, I suppose we'll never know.

 

It's hard, if not impossible, to say.

If they took an FPGA-derived design, and had that fabbed as an ASIC along with other company's designs based on the same raw FPGA/underlying fabric, it could be a lot less expensive than typical "raw" ASICs. Though generally in those cases it's just easier and less expensive to stick with the FPGA.

One would need more insight as to exactly what that IC is responsible for and the parameters of it's operation to be able to make an informed guess one way or another.

 

I'm not informed enough to follow the conversation, I know their past DAC had an ASIC (they called it "musIC" chip) and the Atlantis has a more powerful one (which they call "musIC 2")

Their website explains it a bit:
http://www.wadax.eu/?q=atlantis/dac
I've found two posts where the designer ("Javier_G") explains a bit more about it:
http://www.whatsbestforum.com/showt...-Atlantis-Line&p=457604&viewfull=1#post457604

They're using "45nm" and the DAC is firmware upgradable via internet according to the designer.

 

Then my, entirely-non-researched, opinion would be that it's a hybrid chip. Probably a high-speed, wide register, DSP with a nice wide, fast, I/O controller, based on licensed DSP and vector cores. Which would fit into the "modified design of a large-run chip". Doesn't make it any less special in terms of capability ... just "less expensive"* to have fabbed, especially at 45nm.

--

*Depending on sales volume, this could still be thousands per IC.

 

Yeah, the price is nuts but it's refreshing to see that they put money and research into the design instead of putting some off the shelf chips and random but good looking tubes while charging the same price like some DAC makers do.

 

Yggdrasil w/ production Analog 2, USB Gen 5 and DSP updates on deck.

Once that's had some warm-up time I can compare it to the pre-prod Analog 2 boards and see if they sound the same. If so, the "Analog 2" impressions/comparisons will be forthcoming shortly there after. If not, then it'll be a week(ish) before I can say anything about them.

 

I think something is missing ...

 

What about comparing to your preproduction Analog 2 Gungnir Multibit??


*Secretly hoping @Torq slips up and lets the cat out of the bag*