• Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Streamer vs Computer

I reckon that there are parallels between computer spec/requirements and those for amplifiers. It does not make much sense to have either way over-specced, as the power usage at idle is often much higher than necessary. Folks that are using gaming PCs/GPUs for audio, or with amplifiers that are waay over specced tend to waste a lot of unnecessary power.
Obviously you want enough power, but excessive is often wasteful due to idle usage. Why I now have several different specs of both amplifiers and PCs.
Pi's, MiniPCs are perfect for both streamers and endpoints. More DSP options available than dedicated streamers too. Power consumption is definitely more of an issue than it was 5 years ago. Horses for courses.
 
Last edited:
TBH I do not understand the difference.

One is intrinsic to the processor (i.e. it tries parallel execution of the cpu instruction set) and the other is external to this and contained in the application software (i.e. a parallelized app..which was my development bag for 20 years)

An example of the later is Oracle Parallel Query... you run an atomic query (cause oracle guarantees that when the query starts the result set will only show row values at that point and not later updates).... it takes the query and spins off multiple agents working on different parts of table(s) involved (based on rowid) and then once all agents have finished, the individual result sets are returned to the parent and a final sort is done (if needed).

This means all cpu's could be active/over subscribed (as well as all disk i/o channels [relative to where the row data is stored])... depending on the degree of parallelism... but many companies use "watchdogs" that either block parallel queries above a certain degree of parallelism or calculate the cost of the query before it runs and blocks it if it's too expensive.

This stops Karen (or Bob) in Accounts from doing something stupid.

And by pipelining I wasnt referring to cpu pipelining (again my bad for not being more explicit) but as an example, if a music player uses say sox and ffmpeg, you will have sox running on one core doing resampling and ffmpeg on another with the data from sox piped (at the OS level) into ffmpeg

Peter
 
Last edited:
Here is a screen capture of my m4 macmini pro’s activity monitor while playing a song within Apple Music.:

View attachment 433240

Or an even simpler view - the load on the cpu:
View attachment 433241
Most cores are doing nothing.

Playing back audio is not even a blip to the cpu.
agree... that wasn't what I asked...I know all that...I have benchmarked many playback apps (cpu, memory used, network utilization, context switches, disk i/o etc etc)

My question was is there is any music playback software that runs in parallel (see my Oracle example above) to justify using servers for music playback where those servers could run a business. For example is PCM -> massive DSD resolution done in parallel (by my definition).

My mistake originally (which I corrected myself later) is that multi-threading (aka parallel processing) to a software guy (like me) is different to what others may understand multi-threading to be (i.e. cpu threads)

Peter
 
Last edited:
My question was is there is any music playback software that runs in parallel (see my Oracle example above) to justify using servers for music playback where those servers could run a business. For example is PCM -> massive DSD resolution done in parallel (by my definition).
Seems odd request to me. Playback software sits on end user computers not on servers. You can centralize music files and have the distrubuted playerback software pulling from a common source. You need to give more details on the use case as we are just guessing what you are trying to do.
 
My question was is there is any music playback software that runs in parallel (see my Oracle example above) to justify using servers for music playback where those servers could run a business. For example is PCM -> massive DSD resolution done in parallel (by my definition).

Look at the age of various formats.
  • aiff - 1985
  • wav - 1992
  • mp3 - 1991
  • aac - 1997
  • flac - 2001
  • alac - 2004

Most of the formats and algorithms designed to process them where designed with low end consumer hardware in mind. Hardware has moved on a lot sine most of these formats came into existence, so you can do far more now than you could back then.

As someone else mentioned a $15 Pi Zero 2 W will happily handle just about anything you might need for streaming. With that being said, you average consumer isn't super tech savvy and would likely shy away from something as minimalistic as a pi. Hence why a lot of people go with mini pcs, they are a concept they are more familiar with.

I purchased 2 of these to make receptionist computers for my wife's business, and even they are over kill for streaming music to a person or 2.

The people I know who are using anything close to a server, aren't just streaming audio. They are running TrueNas or something similar ,and because it supports plugins/apps that lets them set up a lot of stuff on one machine, assuming it's powerful enough.
 
Last edited:
agree... that wasn't what I asked...I know all that...I have benchmarked many playback apps (cpu, memory used, network utilization, context switches, disk i/o etc etc)

My question was is there is any music playback software that runs in parallel (see my Oracle example above) to justify using servers for music playback where those servers could run a business. For example is PCM -> massive DSD resolution done in parallel (by my definition).

My mistake originally (which I corrected myself later) is that multi-threading (aka parallel processing) to a software guy (like me) is different to what others may understand multi-threading to be (i.e. cpu threads)

Peter

Why would you want to run your music on a buisness server (as I presume Oracle is for buisness since I don't know many people running it at home for fun and I'm a programmer more than 40 years).
Having said that, you can run ROON server to do your upsampling/trancoding using HQPlayer and you can fiddle with Threads and CUDA loading of the process (HQPLayer supports that) and have an enpoint supporting ROON via its propriatary network protocol or DLNA (which limits your bitrate).

Roon Networking Best Practices
Best Practise for Roon - Networking - Roon Software Discussion / Networking - Roon Labs Community

Having said that I hate ROON since it sucks big time in usability and more.

Beside ROON you can use FOOBAR2000 via DLNA and have MonkeyMote installed on tablet to control it.
Best option IMHO,
 
My vote:

Foobar2000 + MathAudio RoomEQ VST + EasyQ VST

Minimserver

Hifi Cast

All free with excellent functionality. Plus use minimal resources.
 
Last edited:
Back
Top Bottom