How important is hardware for UPnP server duties?

I would like to the obtain the best possible SQ from a UPnP/DLNA server, but have heard conflicting opinions on how important the hardware side of things is when it comes to such duties.

There are a few options for a UPnP server:

1. NAS with or without linear PSU

2. NUC in a fanless case, with or without linear PSU

3. Raspberry Pi, with or without linear PSU

Since the signal sent over UPnP/DLNA protocol is data only (please correct me if I’m wrong), different rules should apply. Many people say the hardware choice for such a server is irrelevant as long as it works, as it shouldn’t or won’t affect sound quality, especially if the signal travels through an optimized network path (fiber optical isolation, OCXO switches, lan filters, etc.). 
 

What do you think? Should I invest in expensive equipment (say a dedicated audiophile server) or is a simple PI4 enough? How important is linear PSU there?
 

11 minutes ago, The Computer Audiophile said:

 

As you already know, the opinions on this will range all along the continuum from "can't possibly matter" to "it's critical."

 

Yeah, I would like to hear some pros and cons from both schools of thought! ☺️

I’ve played around with different options over the years for where to pull music from, local and remote storage. NAS storage always worked best regardless of my player solution. SMB, UPNP both work well. Power supply noise only affected down stream devices so an Etherregen was a more efficient way to protect the player/endpoint solution devices from noise injected into the network. The only lack  I feel with an atom powered NAS is the lack of CPU horsepower when a library refresh is required by the player.

I love playing around with different methods of building a music server.  From my Synology NAS to my primary MAC desktop that has 30Tb of Direct Attached Storage.   My current favorite is a Raspberry Pi 5 with 4TB of NVME storage.  I run headless Pi OS and both LMS and Minimserver.  I treat the music server as a stand-alone appliance.  It does one thing: source music. 

5 hours ago, bobfa said:

I love playing around with different methods of building a music server.  From my Synology NAS to my primary MAC desktop that has 30Tb of Direct Attached Storage.   My current favorite is a Raspberry Pi 5 with 4TB of NVME storage.  I run headless Pi OS and both LMS and Minimserver.  I treat the music server as a stand-alone appliance.  It does one thing: source music.

So you went from a NAS to a MAC and settled on a PI5. Can you elaborate why you took such a path?

For example, your opinion on the following:

NAS powered via its supplied switching PSU in a separate room

                                      vs.

Raspberry PI (or some audiophile versions of it) powered by a good linear PSU in the listening room 

I actually have the NAS and the MAC and the Pi Running at the same time to bounce between them.  The mac and the NAS are on the main lan and the pi 5 is on the audio lan.

 I run fiber from the main network switch to a dumb switch with one fiber connection.  That gives me four ethernet ports for the music systems behind their own switch.  That switch is on an LPS.   

Lots to write about.

On 8/15/2024 at 10:40 PM, AmusedToD said:

Since the signal sent over UPnP/DLNA protocol is data only (please correct me if I’m wrong), different rules should apply. Many people say the hardware choice for such a server is irrelevant as long as it works, as it shouldn’t or won’t affect sound quality, especially if the signal travels through an optimized network path (fiber optical isolation, OCXO switches, lan filters, etc.). 
 

What do you think? Should I invest in expensive equipment (say a dedicated audiophile server) or is a simple PI4 enough? How important is linear PSU there?

I've experimented with a lot of different hardware configurations over the years...since 2001 or so....

Back then, around 2001 I used a computer with a ATX form factor using WindowsXP with Foobar2000-Otachan ASIO plugin and Winamp. My hardware tweaking was trying to find the best RAM I could so I leaned towards the same RAM gamers used and tried to go fanless. My mind told me there were some improvements, mostly gained with better RAM.

Later I moved on to mini-ITX form factor motherboards and experimented with different CPU's, CPU tuning and RAM tuning along with better power supplies. Again RAM provided the biggest improvement in sound. I also learned the changes in CPU speed and function changed the sound but not always for the better. Using a linear power supply on my mini-ITX board provided a noticeable but small improvement, however it was consistent with every mini-ITX board I tried it on, that would be darker background, tighter bass and less glare.

When SBC's came on the scene I made it a point to try as many as possible. So BB-Black, Asus, RPi, Odroid to name a few were all on the table. By this time I knew what to expect from power supplies and had a pretty good understanding that RAM eg to speed of data transmission and the absence of jitter in that data transmission played a big role in better sound. I also A-B'd SBC's with other larger motherboards I had on hand [ATX form factor and a few different mini-ITX boards] All using the same OS and the same software. Over the course of nearly 3 years of weekend fun I came to learn processor power over 2Ghz was not needed, RAM requirements were proportional to the OS workload, power supplies always provided solid but med-small gains and less is more....always. 

So much like getting the best from tube gear, its a combination of adjustments, improvements and tweaks on different fronts....a balancing act, like a house of cards, that provides the best gains. First and foremost listen too and trust your ears, greater cost does not equal greater sound and try not to go down the rabbit hole of diminishing returns.

These days, I've settled on standard the best affordable power supply, the smallest and simplest OS possible preferably with a single purpose or function, utilize RAM to the fullest, and use the least amount of peripheral gadgets as possible.   

On 8/15/2024 at 9:40 AM, AmusedToD said:

I would like to the obtain the best possible SQ from a UPnP/DLNA server, but have heard conflicting opinions on how important the hardware side of things is when it comes to such duties.

 

There are a few options for a UPnP server:

 

1. NAS with or without linear PSU

2. NUC in a fanless case, with or without linear PSU

3. Raspberry Pi, with or without linear PSU

 

Since the signal sent over UPnP/DLNA protocol is data only (please correct me if I’m wrong), different rules should apply. Many people say the hardware choice for such a server is irrelevant as long as it works, as it shouldn’t or won’t affect sound quality, especially if the signal travels through an optimized network path (fiber optical isolation, OCXO switches, lan filters, etc.). 
 

What do you think? Should I invest in expensive equipment (say a dedicated audiophile server) or is a simple PI4 enough? How important is linear PSU there?
 

 

 

 

I use a raspberry pi 4 running upclimpd, works like a charm. Moode audio also works great.

i use a CUI medical grade low noise switching power supply recommended by Miska.

My conclusions so far go in favor of RPI4 powered by a linear PSU (Gentooplayer with Audirvāna) in the listening room vs. super powerful PC (Ryzen 7950X3D) in a separate room. 
 

However, RPI4 is pretty much struggling with running Audirvāna smoothly, so I am starting to think about getting a 13th gen NUC and a 19v linear PSU for the purpose of running Audirvāna core.

Typical UPnP servers (like Minim and Asset) don’t come close to Audirvāna via DLNA in terms of SQ.

I am pretty sure Tidal and Quboz do not run on audio grade servers. Most important for Upnp and DLNA is the network path just like one will optimize the network path for streaming Tidal and Quboz.