Why does the computer matter?

[Part-Time Audiophile]

Sorry to be dense -- but do you prefer Windows to Mac OS? Seems you're favoring Windows? Not that there's anything wrong with that. ;-)

 

RE: "Jitter" in the computer -- I'm not sure what you call all those goings on in the computer that could affect the timing of data delivery, but I call that stuttered-send "jitter", for lack of a better term. In fact, if there's nothing in the computer telling it that it needs to treat this data stream differently from any other, there's no reason to think the other processes, also competing for resources, will care. That's why QoS queues are so important in networking -- or why "hog mode" works well, why re-prioritization of critical process threads is probably a good idea, and why an unencumbered computer is probably better than a well-used, multipurpose one. Yes, I know "jitter" has a technical use and I'm abusing it, but I really don't know what else to call "irregular timing" but jitter.

 

[ziggyzack]

'You have a point, however once you move the jitter-sensitive conversion electronics outside the computer, then it is just data handling. Any computer in theory should work perfectly.

 

The reality is that there is CODEC and DSP software involved in the computer, and like most software it is never perfect. This is even the case on Mac. Amarra and Pure Music prove that either the data handling and rounding/dithering is not handled properly in iTunes or the CODECs have flaws, or both.

 

So, even if you build a fortress of a good desktop computer with low noise and good power supplies, it will still have potential software issues.

 

The cheap laptop or Mac Mini does not have errors, so the fortress desktop is a waste of money IMO. In fact, most of the off-the-shelf music servers have issues with ripping software or playback software or both.'

 

Are you saying that, if you move anything jitter-sensitive out of the computer, then any old computer will do (no need to attempt to bring it up to audio-spec), although you're still left with the possibility of buggy software?

 

[PeterSt]

but I call that stuttered-send "jitter", for lack of a better term.

 

... but I don't think this can be the case, because at the other end it does nothing. As long as buffers don't run empty (or overflow) that is.

 

But how I call it since 3 or 4 days is noise impeeded jitter. So, real jitter at the other end, but impeeded by current draw hence (irregular) noise at the computer end. This noise next is able to siple through all the way to the DAC, and no async or anything will help *that*. Only complete galvanic isolation will, but personally I couldn't find any means of this not creating jitter in the first place.

 

It may be a good idea to think about this noise impeeded jitter, already because I suddenly found it just sounds like jitter (when the influence of the software is well overdone (and which I can incur for indeed)).

 

Peter

 

[audioengr]

ZiggyZack wrote:

"Are you saying that, if you move anything jitter-sensitive out of the computer, then any old computer will do (no need to attempt to bring it up to audio-spec), although you're still left with the possibility of buggy software?"

 

Exactly.

 

The goal as a computeraudiophile is to eliminate all of those software issues. Some things that seem to help on the computer are:

 

1) SSD for the OS

2) lots of RAM

3) run 64-bit mode

4) custom playback software - Amarra or Pure Music

5) careful attention to ripping drive and software

6) if its a PC, use Kernel Streaming or WASAPI

7) kill all unnecessary background applications

 

I dont know why these seem to help, but I have lots of reports that they do.

 

Steve N.

Empirical Audio

 

 

[audioengr]

"but I call that stuttered-send "jitter", for lack of a better term."

 

This is not jitter, these are drop-outs due to latency in the computer software or I/O bus.

 

Jitter is this:

 

http://www.positive-feedback.com/Issue43/jitter.htm

 

Steve N.

Empirical Audio

 

[ziggyzack]

Thanks Steve,

 

Useful list, although I'm not sure how using an SSD or lots of RAM addresses software issues!

 

I will try to set that list against a 2 box UPnP system, where the computer only has the function of data server, and the player does the decoding and analogue conversion:

 

For the computer

 

1) SSD for the OS - if you like, but no audio impact

2) lots of RAM - if you like, but no audio impact

3) run 64-bit mode - if you like, but no audio impact

4) custom playback software - not required

5) careful attention to ripping drive and software - absolutely

6) if its a PC, use Kernel Streaming or WASAPI - not required

7) kill all unnecessary background application - not required

 

For the player

 

1) SSD for the OS - no large storage required

2) lots of RAM - no large storage required, just enough for a buffer

3) run 64-bit mode - n/a, will be a custom OS

4) custom playback software - yes, bespoke

5) careful attention to ripping drive and software - n/a

6) if its a PC, use Kernel Streaming or WASAPI - n/a

7) kill all unnecessary background application - already minimalist by design

 

ZZ

 

[Wavelength]

Scot,

 

Yes I prefer laptops to desk tops.

 

Sorry to be dense -- but do you prefer Windows to Mac OS? Seems you're favoring Windows? Not that there's anything wrong with that. ;-)

 

You are not reading this very well. OSX is a significant upgrade over Windows.

 

The noise I am talking about is not something you can hear. The noise is the EMI/RFI noise generated by the computer. The iMac is the worst at this. Because of the screen and the such it actually acts like a big microwave dish dumping radiation to everything in approximately 6 feet radius, especially front to back.

 

~~~~~~

 

Guys anything to make the os run more effeciently will make the music sound better.

 

Thanks

Gordon

 

[ziggyzack]

Guys anything to make the os run more effeciently will make the music sound better.

 

This is not a vote of confidence in mainstream, multipurpose, consumer OSs, or computers for that matter.

 

Surely the logical conclusion of this statement is to stop shoehorning computers (as in PCs) into hi-fi. Let the computers serve just data, converting formats on the fly if you like, and keep the audio processing and analogue conversion strictly separate from the server.

 

This doesn't entirely solve the problem: wherever you have a DAC chip, you have digital and analogue stages in close proximity. For this, proper hi-fi power supplies, components, manufacturing processes, layout and casework are the answers.

 

The vast effort and expense people are going to, with SSDs, vast lumps of memory, special software to ensure things are played at native rates - still leaves them with suboptimal music players. Never mind that the quality can still be very good, it is ultimately limited by virtue of starting out as a PC. I do not mean to criticise individual players, but do have a very strong sense that these tweaks are seen as the answer, and that the overall architecture is not being properly challenged. I fear this will result in the popularisation of more and more PC tweaks, instead of an improved architecture.

 

I'm not even calling for keeping servers out of the listening room in this, although my preference is to keep them out. I'm really thinking only of the futility of turning PCs into something they're not. Something must be wrong if your server's OS is a factor in determining audio quality.

 

ZZ

 

[manisandher]

"OSX is a significant upgrade over Windows."

 

Gordon, those who've heard the Mykerinos/Pyramix combo running on a Windows platform might disagree with you...

 

Mani.

 

[The Computer Audiophile]

Hi ZZ - Your clear over the top love for Ethernet based players is fine, but it would be nice of you researched the negative aspects of network based playback as well. This would make your statements much more believable for people who aren't in your camp and it would educate everyone at the same time. Just claiming that nothing matters on the computer side when using ethernet doesn't really help.

 

For example have you looked into technologies like AVnu and Ravena to see why they've been created and the differences between them and UPnP/DLNA? There are issues with UPnP that may be addressed with these more robust technologies.

 

[Part-Time Audiophile]

Appreciate the feedback.

 

The problem with doing any of this (ie, building music servers) is that the results are what they are and you get are what you get -- that is, you have precious little to compare your new baseline to. Its not until you build quite a few of them that certain perhaps non-obvious regularities might become apparent. So, thanks again for the insights.

 

I take it that the screen on a laptop also radiates back to front -- which is problematic, but probably not as problematic as what you get with an iMac. And you can turn off the monitor, which is a benefit (but I suppose you could do the same with the iMac).

 

The new 13" Macbook Pro has ~10 hour battery life and Microcenter has them right now for $999. Add a 50GB SSD ($130), a 2TB WD FW800 drive ($190), and 8GB of RAM ($125) and I think we have a winner.

 

[Wavelength]

Chris,

 

right behind you....

 

UPNP/DLNA from what the companies who make the renders are saying is not going well for them or their customers.

 

The biggest problem is there is no oversight in the protocol so from one to the other the server may work or the control may work but the rendered does not.

 

Also if you think about it of all the streaming protocols Ethernet really sucks. Considering packet size and the ip protocol even if you had 1GHZ devices you still would only be getting like USB HS/Firewire 400 speeds. If you had any 100Mhz devices well then... consider it less than Full Speed USB capable.

 

I have a Buffalo server that has DLNA and it sucks. Its better with iTunes but some of the renders won't talk at all and some of the control see a on the 24 bit stuff for some reason and none of the 16 bit red book... and they are all the same file format???

 

~~~~~~~

 

Mani... as for Windows, come on really do I have to listen to this.

 

Look I have designed built and shipped more PC's than anyone on this board. I have source code to freaken ton of PC stuff. I personally have released 135 commercial software titles on Windows. I have 64 PCI hardware products which shipped more than 10M. I have designed 5 mother boards that were all custom at the tune of 5.4M units shipped. I wrote most of the custom BIOS's in assembler.

 

I would be the first to say Windows works better than MAC.

 

The point is... it just does not even come close.

 

Thanks

Gordon

 

[PeterSt]

Mani... as for Windows, come on really do I have to listen to this.

 

Since you, as an engineer, are one of the last convicts, I guess you have to, yes.

 

Please don't act if you knew it all right along. Others do, and now you are just following.

 

I really don't want to be rude, but I guess you should not be too.

 

I didn't say you are wrong, nor do I say you are right. I only say you don't own the wisdom. I hope this is fine with you and everyone.

 

Regards,

Peter

 

[4est]

Maybe you ought to pre package some servers to sell w/ or w/o your DACs. I own a PC, but never bothered to configure it for music.

 

How much are the Merging Tech cards anyway?

 

 

 

[kana813]

Current set up: WD HD/WNDR3700-Cat5-PSAudioPWD/Bridge/tagNplay app on iPad.

 

Other than the Player sometimes not appearing on the tagNplay

app,everything works great. Will probably get better when PSAudio releases their tagNplay Server.

 

[Miska]

I would be the first to say Windows works better than MAC.

 

And then there's Linux, which beats both...

 

But seriously, with proper hardware any of these operating systems can do just fine. As long as audio interface hardware is doing busmaster-DMA it doesn't matter what kind of OS there is.

 

 

[manisandher]

... Peter, I was just about to write that the only reason I'm still on a Windows platform is because of XXHighEnd!

 

Gordon, you know I really couldn't care less which OS I use. I'm not a fan of Windows and would be really happy to use OSX... if it translated into better sound. And this is the only criterion I have. By your own measure, we should all switch to Linux. And I would do so if it improved the sound. However, some people I know and trust have been there and not liked the results.

 

From my own experience, and listening to that of others, I think it's fair to say that it's easier to tweak OSX to sound good than it is to tweak Windows to sound good. But both can sound good. Can we just agree on this and leave it there?

 

Mani.

 

[Ned Kelly]

Hi All,

 

Interesting thread, and its great to see so many people looking for answers inside the noisy, difficult environment that is the case of a desktop computer. I'd like to move away from specific problems like RFI and jitter and pose this question - where would we be if computer audio had undergone the same radical progression that video has over the last 10-15 years ?

 

A different set of problems - no question - but I have little doubt that the big difference lies in the fact that many gamers will part with 500 to a thousand dollars every 12 or so months to be able to play the latest games at their highest settings. If NVIDIA and ATI didn't have that massive market, they couldn't hope to recoup their development costs and they definitely couldn't offer a wide range of cards with each new 'evolution' in their GPUs - gamers at all levels benefit from the trickle-down effect : even if you have a crippled version of the 'card everyone wants', you can (usually) still enjoy the same games as your friends. Even the console gamers have enormously powerful hardware at their disposal, and a legion of developers who work 24/7 to get a slice of a very lucrative market. Google the stats on 'Call of Duty : Black Ops' sales figures if you need confirmation that people are making serious money from the video output of PCs and televisions.

 

Enter audio, stage left. Those same guys who spend up big on motherboards, CPUs and video cards seem to dedicate a tiny portion of that figure to getting the best audio experience from their games - thats a generalisation, and I know that some will be using HT kit etc, but many seem content with whatever Logitech gear they can afford : some will still shell out for a dedicated soundcard, but its not the same market that it was in the 90s. Sure, there is an 'audiophile' market, be they music creators or simply consumers, and there is a pro market, but how many of the big PC manufacturers have invested in the development of dedicated soundcards ? Asus - yes, with the Essence, but the rest are happy to have their marketing department include the 'audiophile' tag in any laptop promo where they spent $10 on a pair of Harmon-Kardon speakers. Intel made a point of upgrading the aging onboard graphics for the Core i3/5/7, but its largely a poor second cousin to video because that is what grabs you in the showroom. Ask the average punter whether he would prefer bit-perfect audio or 3D graphics ...

 

I apologise for the length of this post, but I believe absolutely that all of the problems mentioned in this thread, regardless of their actual impact on SQ, can be addressed with enough time and money. Whether the average consumer is prepared to pay a thousand dollars to have onboard sound on par with their killer video card is a tougher question. I agree with earlier posts that point to how cheaply 'consumer' PCs are specced, and I deal with component failure on a weekly basis - a $500 PC is a disposable electro-toy, but how many of us feel the same way about a 5K Mac Pro or similar PC-based server, and which of these will still be worth upgrading in 3 years time ?

 

Ned

 

[The Computer Audiophile]

Thanks for the commentary Ned. Very interesting points about the gaming industry and consumers' willingness to spend thousands on gaming computers with frequent upgrades.

 

[Ned Kelly]

Chris, I'm sure that many here will argue that 'dreams are free, but this is reality in 2010', and I accept that. You mentioned in your article on the C.A.P.S. server that you felt there was a market for canned Linux music servers - I'd take a different tack on that, and argue that there is a market for a specialised case which makes it easier to isolate the noisiest components from your audio hardware - something beyond the smaller form factor of HTPC cases.

 

Many have discussed the need for galvanic shielding and linear power supplies, so why not integrate all of that into 3 case offerings - small/medium/large ? Allow for extra cost options like sound deadening applied to the inside of the case but leave the choice of motherboard/CP/OS to the system builder. Gamers have so many case options available to them - I believe its our turn now

 

[ziggyzack]

UPNP/DLNA from what the companies who make the renders are saying is not going well for them or their customers.

 

Not from where I'm standing. PS Audio are getting their's together. Linn are going great guns. Naim are extending their range.

 

Also if you think about it of all the streaming protocols Ethernet really sucks. Considering packet size and the ip protocol even if you had 1GHZ devices you still would only be getting like USB HS/Firewire 400 speeds. If you had any 100Mhz devices well then... consider it less than Full Speed USB capable.

 

This is remarkably misinformed. Gigabit ethernet (not GHz) has a throughput of around 80 MB/s after network overhead. Fast Ethernet, which has a data rate of 100Mbits/s (not 100MHz) has a real-world throughput of around 8MB/s. The data rate of a stereo 24/192 stream is just shy of 1MB/s. Typically, stream players use a Fast Ethernet port, as this provides ample headroom without the higher power requirements of a Gigabit port.

 

For info, DLNA is widely regarded as a poor standard, poorly conceived and with poor certification. It's far more of a hassle with video than with audio, but still has a generally bad reputation. It's also completely unnecessary, as there is UPnP. The players I mentioned function using just UPnP, which is a well-defined standard. There is a large eco-system of UPnP renderers, control points and servers which all work together as specified. The same cannot be said of DLNA-certified devices and you are wrong to conflate them.

 

ZZ

 

[ziggyzack]

I don't see why firewire players cannot be made in a similar way, I have no particular ethernet fetish, but I do have an architecture fetish, and any architecture where the computer matters is really flawed imho.

 

Those new technologies you mention are interesting - well in fact I could only access AVnu. At first glance, it elegantly addresses lip-sync for AV and multi-channel, and quality of service for highly contended networks. Irrespective of the merits of this, and how far away its introduction may be, it's worth pointing out two things:

 

1) UPnP is not tied to ethernet, you could have UPnP over USB, or firewire, or possibly AVnu if you wished. All UPnP does is describe an architecture with three types of things, a renderer, a server and a control point, and the interactions between them. It is format agnostic. It's quite abstract. (Unlike DLNA which is a dog's dinner.)

 

2) The title says ethernet is not my point. Well, UPnP is not my point either. Just keep the main server computer separate from the audio parts is my point. SSDs, RAM - these are hardly valves and cartridges. Sorry for the cliche, but tinkering with these components is rearranging the deck chairs on the titanic.

 

ZZ

 

[1audio]

Except for small dedicated platforms (ITX types) linear power supplies are not going to be practical for PC's. The minimum supply for a PC today is around 220W. A linear regulated power supply will at best be close to 50% efficient. If you look at the sized of a 250 watt mono amp you will get a sense of the requirements and the heat output. I don't think you can ship a consumer product in the EU with such a supply since its so ungreen as to be brown. Chris launched me on a project like this some time ago and we changed directions once it was all on paper.

 

Ethernet throughput for audio and video:

I have done a lot of testing on Ethernet throughput. I usually can get 95Mbps+ on wired 100t Ethernet using UPD. More than enough for an H.264 High Profile BluRay. Wired GB Ethernet seems to top out at 300 Mbps with most switches and hardware available to consumers. Optical stuff can go much faster but is very expensive and clumsy. Powerline networking can deliver as much as 80 Mbps but 50 Mbps is more likely in practice. 802.11n (fully tricked out) may hit 40 Mbps when you are close enough to not need it. However any of these is more than enough to pass HDTV (20 Mbps) let alone high res audio (maybe 12 Mbps for 24/192 stereo). 802.11 Wireless isn't stable enough in my experience for audio or video. Works sometimes and not other times.

 

This a good tool and easy to use for testing throughput between computers: http://www.ixchariot.com/products/datasheets/qcheck.html there are others but they are either expensive ($5K) or command line geek stuff.

 

 

 

[The Computer Audiophile]

Hi Ziggy - I must have missed your comments where you didn't talk about network streamers :~)

 

I'm not sure why you think FireWire could be used but not USB in your ideal system. Try using a FireWire DAC without a good Ethernet card driver. It will raise havoc even though it doesn't seem like it should.

 

Sure UPnP isn't tied to Ethernet but I challenge anyone to name some UPnP devices readily available that don't use Ethernet.

 

I believe Linn, your favorite, uses DLNA.

 

Suggesting that any system where the computer matters is flawed is like saying all amps sound the same and any amp where the tubes matter is flawed. Ideally nothing would matter.

 

Why not just wait for DACs with built-in SDHC cards up to 2TB? It may sound like the panacea but I'm sure they will be issues with these devices as well. There's no free lunch.

 

[Ned Kelly]

Chris, I completely agree re the 'no free lunch', and I'm not sure why we continue to believe in a silver bullet when the history of computing is littered with compromise. Tim Berners-Lee and many other greybeards were aghast when they realised that HTML and TCP/IP would form the basis for the Web, but it was what they had at the time. We inherited the QWERTY keyboard from a typewriter designed to slow people down, and so it goes.

 

I dont discount the theoretical issues being raised here, but I would dearly love to be part of an afternoon of A/B testing where we listened via identical USB DACs out of a battery-powered laptop and from a well designed desktop machine running from mains power. I make no predictions on the outcome, and I know that many audiophiles are vehemently opposed to DBT, but personally I would be at the front of any queue to take part in such a test.