Regardless of who has the superior algorithm, heavy lifting tasks can be performed much more efficiently and with much higher precision on chips and FPGAs. This is a well known fact....

[Superdad]

Okay I'll complete all of my tests using the Mirus instead.

 

So far I haven't been able to match the Mirus/SD card implementation's sound quality using HQ player with connection to Mirus Via USB.

 

I haven't tried the NAA renderer approach yet but I will. However it will still use USB to interface to the DAC.

 

With Ravenna, the Ravenna interface will be the renderer. So this will replace the role of the NAA. It will also be able to output I2S over LVDS direct into the Mirus HDMI port. This will be the ultimate.

 

I'll report my results of the different approaches on this thread.

 

Well the Sonore Signature Rendu (SONORE Signature Series Rendu) has terrific I2S over LVDS (courtesy of J. Swenson), but of course its Ethernet renderer module can't run Signalyst's NAA.

[Blizzard]

Well the Sonore Signature Rendu (SONORE Signature Series Rendu) has terrific I2S over LVDS (courtesy of J. Swenson), but of course its Ethernet renderer module can't run Signalyst's NAA.

 

 

 

It looks like a very nice renderer. And I would be able to plug it right into the HDMI port on the Mirus. However I'm also working on a renderer as well. It will be 8 channel 24/384 PCM/DSD 256 capable. Along with that it will have onboard DSP capabilities for active xovers, room correction etc.

[MikeJazz]

It looks like a very nice renderer. And I would be able to plug it right into the HDMI port on the Mirus. However I'm also working on a renderer as well. It will be 8 channel 24/384 PCM/DSD 256 capable. Along with that it will have onboard DSP capabilities for active xovers, room correction etc.

 

I am starting to be very interested by the concept of renderer.

 

Would be nice to include a "legacy" spdif output.

[MikeJazz]

Well the Sonore Signature Rendu (SONORE Signature Series Rendu) has terrific I2S over LVDS (courtesy of J. Swenson), but of course its Ethernet renderer module can't run Signalyst's NAA.

 

Mayb something to be taken care at e new release of the Rendu?

[Blizzard]

I am starting to be very interested by the concept of renderer.

 

Would be nice to include a "legacy" spdif output.

 

 

My renderer will be for exclusive use with my complete system. It will interface direct to my DAC's via I2S.

[Superdad]

I am starting to be very interested by the concept of renderer.

 

Would be nice to include a "legacy" spdif output.

 

Well the Sonore Signature Rendu happens to have one of the very best S/PDIF outputs on the planet! It is actually the more sophisticated part of that product's design (LVDS I2S is not as hard). Oh yeah, that S/PDIF circuit is also a John Swenson creation.

[Superdad]

Maybe something to be taken care at new release of the Rendu?

 

The Signature Rendu does not have a computer or OS in it.

[Blizzard]

Well the Sonore Signature Rendu happens to have one of the very best S/PDIF outputs on the planet! It is actually the more sophisticated part of that product's design (LVDS I2S is not as hard). Oh yeah, that S/PDIF circuit is also a John Swenson creation.

 

If S/PDIF is implemented in a very careful manner (which I can imagine is on the Rendu) it can almost match a good SD card interface.

 

One of the guys on the engineering team for Resonessence (Who's a insanely passionate audiophile) swears by these old CD transports made by Nakamichi back in the 90's. He says the SPDIF output is very close to the SD transport on the Mirus. Only drawback is redbook only output.

 

He invited me to come by his place in the near future for a demo.

 

So yes besides the sample rate limitations, S/PDIF can be very good.

[Blizzard]

If S/PDIF is implemented in a very careful manner (which I can imagine is on the Rendu) it can almost match a good SD card interface.

 

One of the guys on the engineering team for Resonessence (Who's a insanely passionate audiophile) swears by these old CD transports made by Nakamichi back in the 90's. He says the SPDIF output is very close to the SD transport on the Mirus. Only drawback is redbook only output.

 

He invited me to come by his place in the near future for a demo.

 

So yes besides the sample rate limitations, S/PDIF can be very good.

 

This is the unit here. The Nakamichi 1000 MB

 

Good description on this link:

 

http://www.audiokarma.org/forums/showthread.php?t=212973

[Jud]

I'm not sure of the meaning of efficient here... Given an algo, the only requirement is that the processing power is enough to execute it with the required speed (ie not maxing out the processing power).

 

There are subtleties like intermittent high current draws which can introduce noise but I'm not sure why this would be across the board better on FPGA vs CPU. That would boil down more to power rail layout than the other considerations.

 

So frankly I don't think per-se this is a question that has a unique answer.

 

Having said this, someone building a DAC with an FPGA is likely to choose hardware that is good enough for the algo they have. So programming a much more complex algo when it becomes available might not be feasible.

 

Actually the sound Blizzard likes from the Mirus comes from a Resonessence filter loaded onto a stock ESS SABRE DAC chip, not an FPGA, as I've told him and Miska has tried to hint. There is an FPGA in the Mirus, which is used for a couple of functions like decompressing FLAC on the fly. There are DACs that implement filters in FPGAs, such as the well regarded MSB units.

[Paul R]

The point I'm getting at is that there's more to a complete system than just algorithms. If your goal is ultimate sound quality there's better ways. If your goal is a better user interface there's better ways.

 

Actually, no. There is not a better way than using a computer.

 

You can dress that computer up in a milled aluminum case and charge 10X what it is worth, but it is still just a computer. There is simply no getting around that, and the difference between a $10K "music server" and a $600 Mac Mini is pretty minute. That is also a fact.

 

But to get back to the subject of the thread topic. The reason I'm using the SD card reader for an example is because the FPGA is doing 100% of the processing. There's no CPU in the picture except for the one programmed into the FPGA. In my experience so far that system produces the best sound. So based on this the FPGA is better at doing the heavy lifting. Maybe it's not because of the algorithm, but because of the compromises involved in the process of getting the audio from the computer to the DAC chip.

 

Perhaps a lighter weight algorithm preformed on a FPGA could be better than a heavy weight one on a CPU.

 

Either way I'm not knocking the technology. I think it's great. And I hope he keeps developing it.

 

You then, are simply fooling yourself. Let's put this into English - a FPGA is a CPU that you can program the millicode logic in. As compared to an ASIC, which would have that programming done for you, such as an Intel, ARM, Power, etc. CPU. It is still a computer.

 

I have no idea what marketing fluff has gotten around to make it seem otherwise. It is the same fluff going around that says a "music server" is better than a computer, given that that "music server" is nothing more than a computer in fancy dress.

 

Oh, ithe "music server" also has all the peripherals selected for you, but then, so does a CAPs server or indeed, even a Mac Mini. (shrug)

 

A rose by any other name is still a rose. And misinformation spread around like this still has it's own unique smell too.

[miguelito]

Actually the sound Blizzard likes from the Mirus comes from a Resonessence filter loaded onto a stock ESS SABRE DAC chip, not an FPGA, as I've told him and Miska has tried to hint. There is an FPGA in the Mirus, which is used for a couple of functions like decompressing FLAC on the fly. There are DACs that implement filters in FPGAs, such as the well regarded MSB units.

 

OT a bit...

 

I was a grad student in Physics and the experiment I was on had such high capture rate that the initial event filtering was done with very specialized chips built for us. I'm sure every high energy physics experiment since then has required such technology. I seem to recall the chips were FPGA but of a very specialized nature.

 

... Back to the scheduled programming in your batchannel

[Paul R]

I'm not sure of the meaning of efficient here... Given an algo, the only requirement is that the processing power is enough to execute it with the required speed (ie not maxing out the processing power).

 

There are subtleties like intermittent high current draws which can introduce noise but I'm not sure why this would be across the board better on FPGA vs CPU. That would boil down more to power rail layout than the other considerations.

 

So frankly I don't think per-se this is a question that has a unique answer.

 

Having said this, someone building a DAC with an FPGA is likely to choose hardware that is good enough for the algo they have. So programming a much more complex algo when it becomes available might not be feasible.

 

The other consideration is that any design is going to be cost constrained. The closest we come to that not being true is is small designs like the John and Alex's REGEN. Even there, I am pretty sure there was quite an effort made to keep the cost in line. Companies like Bryston, NAD, even McIntosh and others, all pay attention to the cost of the components and choose components that balance the cost vs. performance.

 

Then some companies mark it up to 5X - 10X or more the cost of manufacture. Both to stay in business and make a bit of profit, and to appeal to people who assume "it costs more, it must be better!"

 

-Paul

[Jud]

OT a bit...

 

I was a grad student in Physics and the experiment I was on had such high capture rate that the initial event filtering was done with very specialized chips built for us. I'm sure every high energy physics experiment since then has required such technology. I seem to recall the chips were FPGA but of a very specialized nature.

 

... Back to the scheduled programming in your batchannel

 

I'm guessing there will be info somewhere on what the LHC is using (MIT Tech Review, maybe?).

[Blizzard]

Actually the sound Blizzard likes from the Mirus comes from a Resonessence filter loaded onto a stock ESS SABRE DAC chip, not an FPGA, as I've told him and Miska has tried to hint. There is an FPGA in the Mirus, which is used for a couple of functions like decompressing FLAC on the fly. There are DACs that implement filters in FPGAs, such as the well regarded MSB units.

 

You are probably right in the case of the Mirus the filters are loaded in the chip. But the same filters could also be applied on the FPGA if the chip didn't have that feature.

[miguelito]

I'm guessing there will be info somewhere on what the LHC is using (MIT Tech Review, maybe?).

 

I would think so. If you're interested I can ask my friends - those who have not sold their souls to Wall St but cannot afford an Ongaku either - about it. Most are working in CERN.

[miguelito]

The other consideration is that any design is going to be cost constrained. The closest we come to that not being true is is small designs like the John and Alex's REGEN. Even there, I am pretty sure there was quite an effort made to keep the cost in line. Companies like Bryston, NAD, even McIntosh and others, all pay attention to the cost of the components and choose components that balance the cost vs. performance.

 

Then some companies mark it up to 5X - 10X or more the cost of manufacture. Both to stay in business and make a bit of profit, and to appeal to people who assume "it costs more, it must be better!"

 

-Paul

 

Well, lets be clear: not every computer is built the same. Better power, more careful layout, etc etc make a diff. What that is worth is not for me to judge - but I do agree it generally seems overpriced.

[Blizzard]

Actually, no. There is not a better way than using a computer.

 

You can dress that computer up in a milled aluminum case and charge 10X what it is worth, but it is still just a computer. There is simply no getting around that, and the difference between a $10K "music server" and a $600 Mac Mini is pretty minute. That is also a fact.

 

 

 

You then, are simply fooling yourself. Let's put this into English - a FPGA is a CPU that you can program the millicode logic in. As compared to an ASIC, which would have that programming done for you, such as an Intel, ARM, Power, etc. CPU. It is still a computer.

 

I have no idea what marketing fluff has gotten around to make it seem otherwise. It is the same fluff going around that says a "music server" is better than a computer, given that that "music server" is nothing more than a computer in fancy dress.

 

Oh, ithe "music server" also has all the peripherals selected for you, but then, so does a CAPs server or indeed, even a Mac Mini. (shrug)

 

A rose by any other name is still a rose. And misinformation spread around like this still has it's own unique smell too.

 

The best sound I'm getting from the Mirus is from the SD card reader. No matter how good the computer is, getting the sound from it to the DAC is where the compromise lies. This is why $2800 renderer's like the Rendu were created.

[YashN]

I'll try it. I have the perfect computer for the task. If I can prove to the Resonessence engineers wrong, Miska might just be on to something. But I still feel that the USB technology will be the bottleneck.

 

With HQ Player, set aside some time to experiment with all the different filters and modulators and settings generally. The related threads have great info from various members.

 

The bottlenecks are similar in all the tech I have seen because of protocol packets as well as that much, much lesser known phenomenon hiding in that DDR3 Flipping Bits thread.

 

Try everything you can try (including USB + USB Cables + Regen), it's worthwhile.

[YashN]

You then, are simply fooling yourself. Let's put this into English - a FPGA is a CPU that you can program the millicode logic in.

 

Paul, that's an odd way to describe it. They are Gate Arrays whose configuration are programmable in the form of a hardware description language. Changing the description in this text actually provides similar results as wiring actual electronic components differently. I think that's closer to a good description.

 

I have no idea what marketing fluff has gotten around to make it seem otherwise. It is the same fluff going around that says a "music server" is better than a computer, given that that "music server" is nothing more than a computer in fancy dress.

 

Oh, ithe "music server" also has all the peripherals selected for you, but then, so does a CAPs server or indeed, even a Mac Mini. (shrug)

 

A rose by any other name is still a rose. And misinformation spread around like this still has it's own unique smell too.

 

Not sure why you think marketing fluff is the issue here. The real issues are in the realm of EMC and power consumption, concomitant noise affecting SQ. It follows that a smaller circuit near the DAC can be better for SQ rather than having a whole computer directly connected to the DAC because the computer being a general purpose one had a much, much larger EMI/RFI footprint, even more so in the case of OSes like Mac OS X where you can't go down to a very deep level for customisation (unlike Linux).

 

This could explain why Blizzard is getting great results from the SD Card with his DAC. Its also why Miska proposes the client-server Ethernet mode with HQ Player and an NAA.

 

A Mac Mini is cool out of the box, but if audiophiles hadn't heard enhancements with some additional changes, there wouldn't be any product to add to it. i believe there are at least two from Uptone Audio and John Swenson and there are many more things you can do to make a Mac Mini sound even better.

[YashN]

The best sound I'm getting from the Mirus is from the SD card reader. No matter how good the computer is, getting the sound from it to the DAC is where the compromise lies.

 

This is why HQ Player (main computer) + Ethernet + Optical isolation + smaller program by Miska (NAA device) -> DAC is what you should try and compare it with.

[Jud]

I would think so. If you're interested I can ask my friends - those who have not sold their souls to Wall St but cannot afford an Ongaku either - about it. Most are working in CERN.

 

I would be very much interested. If you don't want to go OT, feel free to PM.

[jabbr]

OT a bit...

 

I was a grad student in Physics and the experiment I was on had such high capture rate that the initial event filtering was done with very specialized chips built for us. I'm sure every high energy physics experiment since then has required such technology. I seem to recall the chips were FPGA but of a very specialized nature.

 

... Back to the scheduled programming in your batchannel

Of course initial capture is specialized because you can't typically order the requisite sensors on new-egg.

 

Years ago we had to do these things because the general purpose CPU wasn't so powerful. Think when the Cray ran supreme. In the early 1990s I spent/wasted a fair amount of time hand optimizing some deconvolution and 3d rotation algos for this VLIW chip Intel made called the i860 ... oh and then they put the out of order execution/branch prediction logic on the chip, and now the NIC sends data directly to the Xeon cache.

 

Why do you think the 40 Gbs infiniband cards are so cheap these days? ... because 100 Gbs Infiniband has become standard. Exactly how much data were you collecting? I can tell you exactly what these experiments use these days: parallelized data collection/HPC clusters. Typically a "whole lotta Xeon" (da da de-dum, de de, da da de-dum...).

 

Now for some real large scale data collection they tell you how they do it: https://nsa.gov1.info/utah-data-center/ ... xeons and custom asics for interconnects.

 

Discussion of scientific uses: http://cacm.acm.org/magazines/2015/7/188732-exascale-computing-and-big-data/fulltext and http://www.theguardian.com/science/2015/jul/17/the-square-kilometre-array-radio-silence-in-western-australia-for-most-powerful-telescope-in-history

[Paul R]

Well, lets be clear: not every computer is built the same. Better power, more careful layout, etc etc make a diff. What that is worth is not for me to judge - but I do agree it generally seems overpriced.

 

If we had some measure or guidelines to go by, I would agree with you. But most modern computers, even the $350 models from Best Buy, are very good indeed. Far better than their counterparts were ten, or even five years ago. Quiet, both environmentally and electrically, powerful, usually easily upgraded, and for the most part, very very reliable. It was a shock for me to realize just how good they really are!

 

I really begin to suspect that the vast majority of the products offered to Audiophiles as music servers are really not any better than they gear they probably already have. It all seems to come down to how you use what you have...

 

But that is a bit of an outlier viewpoint. Certainly there are folks who disagree.

 

-Paul

[MikeJazz]

Well the Sonore Signature Rendu happens to have one of the very best S/PDIF outputs on the planet! It is actually the more sophisticated part of that product's design (LVDS I2S is not as hard). Oh yeah, that S/PDIF circuit is also a John Swenson creation.

 

Yeah, but not available in Europe/not "CE"...

we need something similar on this side of the atlantic!