Uh oh, I beat my SD card trick; Bypass your Ethernet switch, and make your external drives sound close to RAM disk using an Apple Thunderbolt Ethernet Adaptor and a second network connection.

[alcarp]

Has anyone compared booting from an SD card to booting from an SSD connected to the TB port of a mac mini?

[wisnon]

Jud and S-Dad,

 

Stuff like this:

Arduino Project 5: Digital audio player

 

or Pure Super Audio: New SA-SD card player promises DSD file playback without compromise

or something similar?

[alsterfan]

Dear Uwe:

 

...by booting from my same optimized Mavericks SD card...

 

Dear Alex,

I´m glad that you had the time to answer my question. It will make me be a bit more cautious before making a decision. Up to then I want to try to optimize my iMac (Mavericks). On this road I have gone just a few steps because of lack of knowledge how to do that. Perhaps it would be possible for you to write which programs and services you deactivated and how you managed that.

Up to then I am highly enjoying listening to music from RAM. Thanks a lot for mentioning this playback-possibility as well. Using a SD card to store music was a big step forward but using RAM (in spite of some inconveniences) is a way bigger progress soundwise in my chain than I hoped to expect.

 

Best wishes and good luck with the setting-up of a NFS server.

 

Uwe

[Skeptic]

If you're hearing such a benefit from bypassing the internal drives, why not just use a player like JRiver which supports memory playback?

 

Essentially what happens is that it decodes each track into its own "RAM Disk" and plays from there when you enable the option.

 

Then it doesn't make a difference what your storage medium is.

[Superdad]

Jud and S-Dad,

Stuff like this:

Arduino Project 5: Digital audio player

 

or Pure Super Audio: New SA-SD card player promises DSD file playback without compromise

or something similar?

 

Really? The Arduino is 8-bit, mono, and 19KHz--and it is still a processor. The prototype Logifull showed in 2102m has never been seen again, there is isn't a scrap of mention on its web site, and it also must have some sort of high speed processor and s/w in it.

 

There is no free lunch. Software and a processor are not evil, they just have to be implemented correctly. I am all for the various attempts at creating or using a super-minimal OS or core code to playback music in a pure way--and someone could do it in an FPGA, an Atmel, a Xilinx SoC, or an ARM--it is just not easy, and to do it in an open and extensible way is even harder. As I said before, I doubt there is much of a market for a box that does nothing but play tracks from an SD card--especially since the only way for that box to offer the SQ potential of SD is to also house the DAC. And that is a whole other can or worms. We all want some choice and flexibility.

 

I do think that DAC manufacturers would do well to begin including an SD card slot. It does mean that they then have to think about how to initiate and control playback from them. Most DACs--including most USB-input ones--are not really smart host and control devices. That is kind of the dividing line between the various music servers (both home-rolled and commercial) and a "dumb" DAC. Different functions, different devices, though we are seeing some convergence (Community Squeeze and others).

[phatoldsun]

If you're hearing such a benefit from bypassing the internal drives, why not just use a player like JRiver which supports memory playback?

 

Essentially what happens is that it decodes each track into its own "RAM Disk" and plays from there when you enable the option.

 

Then it doesn't make a difference what your storage medium is.

 

They are using a memory playback player. Audirvana+ is the popular one.

[Audio_ELF]

I do think that DAC manufacturers would do well to begin including an SD card slot. It does mean that they then have to think about how to initiate and control playback from them.

How would you feel if the playback / control was simple: start, stop, skip forward / back? The Naim DAC implemented this for USB sticks and it will just play through the tracks on a memory stick in order.

 

Or ... perhaps we should go back to spinning laser read discs?

[Skeptic]

They are using a memory playback player. Audirvana+ is the popular one.

Well I can't comment on Audirvana's implementation, but if the files are not even being played back from local storage, how can it have an effect on audio quality?

[davidbeinct]

Well I can't comment on Audirvana's implementation, but if the files are not even being played back from local storage, how can it have an effect on audio quality?

 

Memory of where the files used to reside? Water memory is a described phenomena in homeopathy, perhaps it hasn't been studied enough with regard to playback of digitally stored files:

 

Memory of water

[phatoldsun]

Well I can't comment on Audirvana's implementation, but if the files are not even being played back from local storage, how can it have an effect on audio quality?

 

That's the $10.00 question. There is plenty of speculation in this and the other related (lengthy) threads linked in the 1st post if you want to take a journey through them.

[Freann]

The next song is being read into RAM while A+ is playing. So there is file activity, but not continuously. Also A+ is extracting and showing embedded cover art etc.

 

@davidbeinct: you are kidding, right? Right..?

[elcorso]

The next song is being read into RAM while A+ is playing. So there is file activity, but not continuously. Also A+ is extracting and showing embedded cover art etc.

 

@davidbeinct: you are kidding, right? Right..?

 

Of course @davidbeinct is right. There is a lot of 'memories':

 

From elephants, very well known, as in dogs. (By smell).

 

From wood. This is anecdotal, but some musicians believe tan wood from old wood music instruments, and from old wood concert halls, the wood is embedded with music, then his nice SQ.

 

From my wife, than can remember from 20 years ago when I went on a spree and came home home happy, drunk and lipstick painted.

 

And so on...

 

Roch

[wisnon]

Really? The Arduino is 8-bit, mono, and 19KHz--and it is still a processor. The prototype Logifull showed in 2102m has never been seen again, there is isn't a scrap of mention on its web site, and it also must have some sort of high speed processor and s/w in it.

 

There is no free lunch. Software and a processor are not evil, they just have to be implemented correctly. I am all for the various attempts at creating or using a super-minimal OS or core code to playback music in a pure way--and someone could do it in an FPGA, an Atmel, a Xilinx SoC, or an ARM--it is just not easy, and to do it in an open and extensible way is even harder. As I said before, I doubt there is much of a market for a box that does nothing but play tracks from an SD card--especially since the only way for that box to offer the SQ potential of SD is to also house the DAC. And that is a whole other can or worms. We all want some choice and flexibility.

 

I do think that DAC manufacturers would do well to begin including an SD card slot. It does mean that they then have to think about how to initiate and control playback from them. Most DACs--including most USB-input ones--are not really smart host and control devices. That is kind of the dividing line between the various music servers (both home-rolled and commercial) and a "dumb" DAC. Different functions, different devices, though we are seeing some convergence (Community Squeeze and others).

The Resonessence Invicta Dac has SD card playback. The interface is not great though. There is also another Dac I saw doing this (Dutch origin).

 

The polluting PC hardware is evil, not software (which can be selectively shut down).

[Jud]

The Resonessence Invicta Dac has SD card playback. The interface is not great though. There is also another Dac I saw doing this (Dutch origin).

 

The polluting PC hardware is evil, not software (which can be selectively shut down).

 

"Computerless" playback is a marketing term. Every digital playback chain needs hardware and firmware/software. The differences lie in how something sounds, not in the name by which a marketing department decides to call it.

[Jud]

The next song is being read into RAM while A+ is playing. So there is file activity, but not continuously. Also A+ is extracting and showing embedded cover art etc.

 

@davidbeinct: you are kidding, right? Right..?

 

One hopes so. Audio has done more for people's well-being than homeopathy ever has, that's for sure.

[Daudio]

There is no free lunch. Software and a processor are not evil, they just have to be implemented correctly. I am all for the various attempts at creating or using a super-minimal OS or core code to playback music in a pure way

 

Since you speak of correct implementation and minimal OS, why not take a step further towards a Real-time operating system, which after all is the proper environment for a time dependent task like playing digital audio (time perfect).

 

General-purpose multiprocessors and operating systems do lots of different stuff, and digital audio is just another task to be juggled with a bunch of others (I wonder if CoreAudio even has any processor affinity ?). We have long assumed that faster and faster processors would automagicaly handle all those differing priorities for us, but maybe not...

 

Perhaps you or someone could experiment with LynxOS, RTLinux, or QNX ? Does Audirvana run in Linux ? Have they been used in some of the black box servers, like the Olive ?

 

Anyway, just an idea for the mix...

[Jud]

If you're hearing such a benefit from bypassing the internal drives, why not just use a player like JRiver which supports memory playback?

 

Essentially what happens is that it decodes each track into its own "RAM Disk" and plays from there when you enable the option.

 

Then it doesn't make a difference what your storage medium is.

 

PeterSt, the programmer of the memory player XXHighEnd, disagrees with you. He thinks having the music and the player application on RAMdisks makes a favorable difference in the resulting sound.

[Superdad]

The Resonessence Invicta Dac has SD card playback. ...The polluting PC hardware is evil, not software.

 

Thanks for pointing that one out Wisnon. The Invicta is a good example. Obviously they wrote software for it (it drives an HDMI monitor with a menu system, etc.), but it appears they did it all in an FPGA or other SoC processor. That's cool, and is the sort of thing I like to see in a DAC. But one has to like their $5K DAC to get that.

 

The line between what is "polluting PC hardware" and what is clean, scaled down processing is blurring. My point was that some of these tiny computers (Beaglebone, Wandboard, etc.) may not be as "evil" as you think, given proper power supply and isolation, and running an OS stripped of superfluous processes. And such devices offer far greater flexibility in storage, playback, and interface.

 

The fiddling about we are doing with dedicated, minimal machines running OS X, Windows, or variations of Linux/UNIX is just an attempt to see how far we can take the SQ without loosing flexibility. There are plenty of other areas of digital music reproduction that can benefit from further attention. USB can be done better than it is now, and with computer power and s/w we can and should begin to rethink the necessity of DAC chips themselves.

 

That last point is the real benefit of keeping (customized) computers in the equation: With Audirvana Plus (or Amarra, JRiver, XXHigh-End, etc.) I can upsample Redbook in s/w to 352.8KHz (with a ear-tuned iZotope filter in the case of A+) and either bypass a DAC's resource constrained built-in filter, or in my case, turn an NOS PCM1704 into a zero-DSM world-class performer.

Or, with the right hardware I could use Miska's HQPlayer and its custom S-D modulator algorithms and convert all PCM to 256 or 512 DSD before it leaves the computer--and eventually use a DAC that does not have a DAC chip. Wisnon, I know you are a proponent of DSD/SACD, so I would think that the vision of that future would be appealing to you. So let's not burn the computer at the stake just yet.

 

Regards,

ALEX C.

[Jud]

Since you speak of correct implementation and minimal OS, why not take a step further towards a Real-time operating system, which after all is the proper environment for a time dependent task like playing digital audio (time perfect).

 

General-purpose multiprocessors and operating systems do lots of different stuff, and digital audio is just another task to be juggled with a bunch of others (I wonder if CoreAudio even has any processor affinity ?). We have long assumed that faster and faster processors would automagicaly handle all those differing priorities for us, but maybe not...

 

Perhaps you or someone could experiment with LynxOS, RTLinux, or QNX ? Does Audirvana run in Linux ? Have they been used in some of the black box servers, like the Olive ?

 

Anyway, just an idea for the mix...

 

If they're non-Linux (like qnx, which I like a lot), the problem is they tend not to have drivers for necessary stuff like DACs.

 

There are actually instances when one doesn't want real-time response in rendering digital audio, such as any time a buffer is desirable. At least one player (XXHighEnd) uses system response timing as a tunable variable, so it appears the fastest response even exclusive of buffers is not necessarily best for sound quality.

[Superdad]

Since you speak of correct implementation and minimal OS, why not take a step further towards a Real-time operating system, which after all is the proper environment for a time dependent task like playing digital audio (time perfect).

 

Indeed, I would love to try out an RTOS if good player/library management s/w was available for one. Anyone know of such a combo that I could easily install on an i5 Intel Mac mini?

 

What's funny is that Windows CE--available to developers for a multitude of processor architectures--is now a real-time OS, and typically used for embedded applications. I wonder what it for take for PeterSt. or Miska (XXHigh-End and HQPlayer respectively) to get their products running under Windows CE? They might both shudder at the thought, but who knows...

[Miska]

What's funny is that Windows CE--available to developers for a multitude of processor architectures--is now a real-time OS, and typically used for embedded applications. I wonder what it for take for PeterSt. or Miska (XXHigh-End and HQPlayer respectively) to get their products running under Windows CE? They might both shudder at the thought, but who knows...

 

Last time I touched WinCE it was completely different beast with stripped down APIs. So quite big task to port there.

 

RT-patched Linux is as realtime as most others, last time I measured response time it was around 5 µs on low power 1 GHz VIA CPU (older model of these). Most real RT-OS don't do any better than that...

[Jud]

RT-patched Linux is as realtime as most others, last time I measured response time it was around 5 µs on low power 1 GHz VIA CPU (older model of these). Most real RT-OS don't do any better than that...

 

RT-patched Linux - RT kernel only, or userland changes as well?

[Skeptic]

PeterSt, the programmer of the memory player XXHighEnd, disagrees with you. He thinks having the music and the player application on RAMdisks makes a favorable difference in the resulting sound.

But it's already being stored in RAM.

[Jud]

But it's already being stored in RAM.

 

Yep.

 

Look, I'm not an audio engineer, and I certainly don't claim to be infallible, so I don't know exactly how this is supposed to work or even if what I consistently hear is objectively real. Whether it has something to do with double buffering in RAM, the system using less resources or making less electrical noise when pulling files into RAM *from* RAM versus pulling from other types of storage, the fact that you don't have an external HDD with a SMPS connected, or some combination, I don't know. But it consistently sounds better to me, and a guy who codes a memory player who presumably knows more about this than we do says that's not crazy.

[wisnon]

"Computerless" playback is a marketing term. Every digital playback chain needs hardware and firmware/software. The differences lie in how something sounds, not in the name by which a marketing department decides to call it.

 

But Jud, the typical muli-purpose computer is a sea of NOISE. Complete overkill too. That is the point.