Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[pluto]

Having recently commissioned a new audio-source computer, I've been through all the usual processes that one goes through setting up networking, buffer sizes, that kind of stuff. It has to be said that everything works entirely as expected without a sniff of a problem, so my curiosity was piqued regarding the nature of the audio pre-buffering parameter and its soulmate, memory playback. Let me say from the outset that I'm not interested in highly conjectural effects on sound quality¹ but rather, the principles of network engineering.

Turning the pre-buffering option way down low causes the network access to operate, more or less, in tiny bites continuously. Increasing the pre-buffer time causes the network access to happen in bigger chunks, less frequently – entirely as expected. Opting for memory playback, in effect, takes this process to its logical extreme insofar as the system takes one massive bite, and then has no further interaction with the network until necessary. Everything appears to happen as it ought.

So my discussion point is this: In the context of audio playback, do the principles of good network engineering express a preference for "little and often" or taking the biggest possible bite and not accessing the network again, sometimes, for many minutes?

¹ I must state, for the record, that I have never heard any evidence to support the belief that so-called "memory playback" has any audible advantages in a situation where the source data is arriving (be it from a local disk or network) in a timely manner.

[JimH]

You're correct.  What matters is that the network works.  You would hear dropouts if it didn't.

[pluto]

So, philosophically, “little and often” or “all you can eat”?

I would be interested in an experienced network engineer's view of this, as the principle obviously applies to many other aspects of network use in addition to playing audio. Would the answer vary depending on whether the network was copper or wireless?

[JimH]

Same answer. It doesn't matter.

Audio is very light duty for any network.

My experience with networks probably pre-dates your adolescence.

A healthy network has no negative impact on audio quality.

[pluto]

...My experience with networks probably pre-dates your adolescence.

I'm older than I look 

A healthy network has no negative impact on audio quality.

But seriously, I'm asking from the point of view of network reliability and efficiency, which has nothing to do with any conjecture about the effect of the network on audio quality.

[JimH]

If you really want to know a lot about networks, you'll have to go digging on the Internet (and not on audiophile forums).

Like computers, networks are extremely reliable, but there are lots of bad things that can happen to them, so it's not possible to generalize in any useful way.

If you're asking about whether a network likes to move big chunks or little ones, the answer is no.   It doesn't care.  Small chunks have more overhead, but at network speeds, it just doesn't matter.

Remember a 10Mb network?  That's really old, but it still moves 10 million bits per second, or (adding some for overhead) about 1 million bytes per second.   A 3 minute track of FLAC audio is around 30 million bytes, so it would take about 30 seconds to be sent over the network.   The audio software does "one potato, two potato, three potato, four, five potato, six potato, seven potato, more,x of waiting for every one potato of work.  Only DSD audio would be a challenge for a 10Mb network, but you probably have a 100Mb or 1Gb network.

In the end, what matters is whether it works.   If you have dropouts, you have to start looking for the guilty.  Otherwise, you need to enjoy the music.