Jump to content

Recommended Posts

Posted

I have 2 questions

I am looking at buying an integrated with a built in dac/streamer

 

(1) First option is I am considering the Roon Nucleus so I can use Roon. 

 

What is confusing me is whether to use the hard disk within the Roon the hard disk should be housed within the streamer or in a separate component such as a NAS Drive or a more audiophile type like the Lumin L1.

 

It would plug into a network switch which also connects to a DAC (housed in an integrated)

 

what is the best option here for best sound.

 

(2) my other option is using the DAC/streamer housed within an amp.

 

In this case would a NAS drive such as QNAP be ok, or do you recommend an audiophile grade one such as the Lumin L1. or does it not matter at all and any NAS drive would do the same thing?

 

 

your thoughts appreciated

 

thanks in advance

 

 

Posted

If  you go Roon, look for a server (e.g. Nucleus), with SSD for the library.  The reduced latency of a local SSD will sound better than getting music from a NAS.  To reduce box count, the server can even be a NAS (e.g. high end QNAP), with SSDs.

 

What is the Integrated with a DAC/Streamer that you are hoping to buy?  I assume it can work as a Roon end point?   If you are not using Roon,  you will likely get better SQ from a powered USB drive than from a NAS.

 

Regarding your question on whether to go NAS or Lumin L1, the L1 will likely have a better power supply and may deliver better SQ.  The NAS is however is more flexible and can be used for many purposes, and by rest of family.  The SQ of music over Ethernet is likely not as good as music from a local drive, but it can be improved (story for another day).  

Posted (edited)

For Roon usage: Roon prefers music stored inside the Roon Core rather than from NAS.  For Roon Core, get a NUC8i7BEH or i5.

 

For Lumin when not using Roon:

 

Although some users did report better SQ, the fanless L1 was primarily meant for Lumin users who do not prefer the complexity of NAS, or want to have matching aesthetics with Lumin players.  Note that L1 does not support Roon because L1 as a UPnP server does not expose a network file system.

 

If you are considering a NAS, you may use the fanless HS-453DX and run MinimServer on it.  Note that it is below Roon Core minimum requirements.

Edited by wklie
Posted (edited)
On 12/11/2020 at 12:15 PM, Snoopy8 said:

If  you go Roon, look for a server (e.g. Nucleus), with SSD for the library.  The reduced latency of a local SSD will sound better than getting music from a NAS.  To reduce box count, the server can even be a NAS (e.g. high end QNAP), with SSDs.

 

What is the Integrated with a DAC/Streamer that you are hoping to buy?  I assume it can work as a Roon end point?   If you are not using Roon,  you will likely get better SQ from a powered USB drive than from a NAS.

 

Regarding your question on whether to go NAS or Lumin L1, the L1 will likely have a better power supply and may deliver better SQ.  The NAS is however is more flexible and can be used for many purposes, and by rest of family.  The SQ of music over Ethernet is likely not as good as music from a local drive, but it can be improved (story for another day).  

 

Not completely correct. It's a function of how you manage interrupt periodicity rather than outright latency.  Both can be 'made' to sound good. 

 

If you're using something internal, power it independently and keep the media separate. 

 

If you're going external, there's a few things to do for sanity of best performanace. 

 

On 12/11/2020 at 1:31 PM, wklie said:

For Roon usage: Roon prefers music stored inside the Roon Core rather than from NAS.  For Roon Core, get a NUC8i7BEH or i5.

 

Roon doesn't really care whether the disk is local or on a network. Network volumes mount at OS level, if in Linux particularly Roon doesn't know the difference.

Edited by rmpfyf
Posted
1 hour ago, rmpfyf said:

Not completely correct. It's a function of how you manage interrupt periodicity rather than outright latency.  Both can be 'made' to sound good. 

Is this something that can be done easily?  Or is this more into esoteric tweaking ?

Posted
3 hours ago, Snoopy8 said:

Is this something that can be done easily?  Or is this more into esoteric tweaking ?

 

With local media you'll want to cut down on sporadic writes to the main disk... the easiest/lowest-hanging fruit is to disable the swap file. The absolute 'yes' is to run the OS out of memory completely (tricky). As for (music) file reads, there's lots of stuff from how you'd setup a NAS share (if you use it) to playing out of RAM and getting the file prepared before playback - no decompression or transformations (you read a lot of 'WAV vs FLAC though it's more accurately FLAC vs 'whatever format is native to your transport' (which is usually not 16-bit WAV). 

 

The remaining issues are broadly electrically-induced in nature. A PC motherboard power supply (not the thing that plugs into your motherboard) isn't the last word in low-noise, it's particularly adept at flipping voltages very very fast as a CPU or other devices change power states hundreds to thousands of times a second (some of these state changes have relatively high latency, and you can usually limit this in BIOS). There's stuff to do in setting frequencies etc relative to e.g. the CPU power rail configuration, memory power rail config, etc. Ideally you have a CPU with no integrated GPU or one that's powered down, as you can end up with the same chip die needing disparate power requirements that don't have an optimal configuration relative to e.g. CPU speed. Unless you have a motherboard where the +12V and +5V for your M.2 are electrically isolated from your CPU+USB then a SSD on M.2 can be flippin' about large bits of current which make for some degree of audible sadness unless independently powered. That's not to suggest all should rush out and get boutique M.2s, more that if you're tossing up where to put data locally a cheaper SATA SSD on a cheap-ish shielded data cable with independent power is just fine.

 

PCs are designed to be robust against timing imperfection so you don't generally see last-word clocks on motherboards. Some are (much) better than others, less by way of quality and more by means of implementation.

 

Then there's OS design etc...

 

Much has been already stated on network optimisation. It doesn't need megadollar cables IMHO.

 

Much of this is esoteric though most tweaks are free andsome basics are easy... get rid of that swap file, and get into your BIOS enough to limit C states. Run that CPU as low as it'll go and lock it there - depending on your motherboard power rail design it could be quite good.

 

 

Posted (edited)
On 14/11/2020 at 12:44 PM, rmpfyf said:

Roon doesn't really care whether the disk is local or on a network. Network volumes mount at OS level, if in Linux particularly Roon doesn't know the difference.

 

Roon software does not.  Roon Labs the company does.

 

Just relaying Roon CTO's preference.

 

It also causes NAS fail to sleep, and some users are not too happy about that.  Then there are additional support issues caused by the use of NAS, such as obscure SMB options that one need to change on some NAS, or simpler ones such as the smb path syntax.

Edited by wklie
Posted
6 hours ago, wklie said:

 

Roon software does not.  Roon Labs the company does.

 

Just relaying Roon CTO's preference.

 

It also causes NAS fail to sleep, and some users are not too happy about that.  Then there are additional support issues caused by the use of NAS, such as obscure SMB options that one need to change on some NAS, or simpler ones such as the smb path syntax.

 

Run a NAS on SMB and you're going to get problems anywhere.

 

NFS. You missed the bit where I mentioned Linux.

Posted
6 hours ago, rand129678 said:

image.png.6dac72660018b84d2d672ee0d02fbd46.png

 

Let me be clear here.

 

Roon, an application or OS+application that's plays music, being about the lowest bandwidth thing to do in multimedia, is indifferent to where your media is if mounted reliably. Not SMB, AFP or any other sh*tful file system, but a reliable scheme like NFS that's used in many time critical applications the world over.

 

Contrary to their COO'S post you cannot 'isolate' an SSD. Music on a local SSD is not any faster. The 'latency' to get media over a network is neither among the latencies you'd worry about for audible performance. That's laughable.

 

The most you'll do in a heavily optimised system - which Roon Core is not - is drop some processes to support the network. And you'd have to have a very heavily stripped back, optimised system with very little schedule management to do so.

 

But sure, if you want to park logic and independent thought on a shelf over a blog post, you can do that too.

 

 

Posted

 

I have a different approach these days.  If it is sounding good, I don't fuss.  To explain,   I listen to streaming radio a fair amount.   Some of them are  high quality, in fact my most recent discovery is one that uses 24 bit 96kHz lossless flac.     Now, this 24 bit data is coming from Icecast radio storage (probably a big NAS) somewhere on the other side of the world, where hard disk access is competing with hundreds of other radio stations with huge numbers of listeners each.  It reaches me via the wild internet,  backbones in different countries and undersea cables etc etc, countless routers and switches.    And it still sounds extremely good.    So, I don't stress very much about local files and where I store them on my little local LAN. Nearly all my local stored files are 16 bit 44.1 kHz. In practice I have had no troubles with using a old slow local NAS, nor with a local USB hard drive (non-SSD). 

Posted
3 hours ago, rmpfyf said:

But sure, if you want to park logic and independent thought on a shelf over a blog post, you can do that too.

 

 

All noted but it's not just a blog post. If it makes a difference, one can email Danny or Brian (CTO) and get the same advice back.

Posted

I use a NAS - and have no troubles at all. It might be a bit fiddly to setup initially - but once that's done it works fine. The advantage of using a NAS is being able to isolate any hard disk noise from the player and being able to serve up music to any player in the house from a single central location. Latency is a non issue - latency just refers to the delay between sending a request to receiving a response and is measured in milliseconds - so you might have a few milliseconds delay between pressing play and actually hearing music - but after it starts playing. latency is completely irrelevant. I used wired Ethernet - it makes for a more stable  platform over wireless - but use regular Cat7 cable - at 1Gbps it has enough bandwidth headroom for it to be a nonissue.

Posted
1 hour ago, notsobitperfect said:

I use a NAS - and have no troubles at all. It might be a bit fiddly to setup initially - but once that's done it works fine. The advantage of using a NAS is being able to isolate any hard disk noise from the player and being able to serve up music to any player in the house from a single central location. Latency is a non issue - latency just refers to the delay between sending a request to receiving a response and is measured in milliseconds - so you might have a few milliseconds delay between pressing play and actually hearing music - but after it starts playing. latency is completely irrelevant. I used wired Ethernet - it makes for a more stable  platform over wireless - but use regular Cat7 cable - at 1Gbps it has enough bandwidth headroom for it to be a nonissue.

I find pretty much the same as you. Even over my 14 year old install of 35M of cat5 up to my shed.

Posted
1 hour ago, rand129678 said:

 

All noted but it's not just a blog post. If it makes a difference, one can email Danny or Brian (CTO) and get the same advice back.

 

Didn't mean to come off as losing it you, apols.

 

Lack of good thought costs much money otherwise.

  • 3 weeks later...
Posted (edited)
On 16/11/2020 at 9:00 AM, rand129678 said:

 

All noted but it's not just a blog post. If it makes a difference, one can email Danny or Brian (CTO) and get the same advice back.

 

There may be an issue with the reasons given for the outcome  rather than the outcome itself.  If SSD does  sound better  - and it will in many situations, then it means that in that particular scenario  there may be better response and less  jitter. However, many others will attest to having graduated from a single server with SSD to  a dual pc setup  where timing /jitter issues have been sorted  and which allows the rendering player to operate with a very low noise floor.

 

These optimised setups are different to your average household NAS which in most cases will struggle  to compete with a server's  internal SSD for sound quality. 

Edited by TP1
Posted
2 hours ago, TP1 said:

These optimised setups are different to your average household NAS which in most cases will struggle  to compete with a server's  internal SSD for sound quality. 

 

Um, no. Quite the other way around. It really does depend how you've got that NAS setup and how the share is integrated - makes a huge difference. The nice thing about doing this with a NAS is that it really doesn't matter whether it's household grade or otherwise. You're just dealing with packets on a network.

 

There have been some excellent posts around how and why to setup as such. 

 

Getting an internal drive to work in a manner free of jitter is not an insignificant effort. 

Posted
7 hours ago, rmpfyf said:

 

Getting an internal drive to work in a manner free of jitter is not an insignificant effort. 


Perhaps I should have explained further. I was referring to highly optimised servers using an internal SSD and comparing that to taking music via Ethernet from an existing NAS. Most  people I know have gone down the route of server optimisation first 

 

In these situations. The internal SSD has generally sounded better than from the existing NAS since nothing much has been done to the psu or anything in between.

 

The better performing dedicated file servers  which are part of a 2 pc setup  have also been highly optimised . Apart from first rate linear power supplies network enhancing options in both units such as JCat Femto NET cards

go a long way to clean up the music source.

 

But then devices such as EtherRegen can have a significant impact on servers which do not derive music from the network at all, but just use it for control . I can only guess the reason why this is so  but  the bottom line is that  based on what I have observed, you will need something more than your garden variety NAS to  sound better than than an SSD  in a decent server. 
 

I am  in the process of conducting more experiments with my Esoteric N-03T.  So far it’s been outstanding using an attached SSD but I have deferred comparing it  to network source until it can be attached to a state of the art unit - which might be in a week or so. 

Posted
On 11/11/2020 at 6:53 PM, ACAUS said:

I have 2 questions

I am looking at buying an integrated with a built in dac/streamer

 

(1) First option is I am considering the Roon Nucleus so I can use Roon. 

 

What is confusing me is whether to use the hard disk within the Roon the hard disk should be housed within the streamer or in a separate component such as a NAS Drive or a more audiophile type like the Lumin L1.

 

It would plug into a network switch which also connects to a DAC (housed in an integrated)

 

what is the best option here for best sound.

 

(2) my other option is using the DAC/streamer housed within an amp.

 

In this case would a NAS drive such as QNAP be ok, or do you recommend an audiophile grade one such as the Lumin L1. or does it not matter at all and any NAS drive would do the same thing?

 

 

your thoughts appreciated

 

thanks in advance

 

 

My advice? As a previous (and happy) Roon Nucleus owner - for the same money you could get an Innuos ZEN Mini which will have 2TB of onboard disk space, a CD ripper and will even give you digital/analogue inputs IF you really need them. Best of all, a ZEN Mini sounds better than a Nucleus for the same money. Try not to overcomplicate things - this is a very good solution that would be easy to sell later on if you get the upgrade bug.

Posted
9 hours ago, TP1 said:


Perhaps I should have explained further. I was referring to highly optimised servers using an internal SSD and comparing that to taking music via Ethernet from an existing NAS. Most  people I know have gone down the route of server optimisation first 

 

In these situations. The internal SSD has generally sounded better than from the existing NAS since nothing much has been done to the psu or anything in between.

 

The better performing dedicated file servers  which are part of a 2 pc setup  have also been highly optimised . Apart from first rate linear power supplies network enhancing options in both units such as JCat Femto NET cards

go a long way to clean up the music source.

 

But then devices such as EtherRegen can have a significant impact on servers which do not derive music from the network at all, but just use it for control . I can only guess the reason why this is so  but  the bottom line is that  based on what I have observed, you will need something more than your garden variety NAS to  sound better than than an SSD  in a decent server. 
 

I am  in the process of conducting more experiments with my Esoteric N-03T.  So far it’s been outstanding using an attached SSD but I have deferred comparing it  to network source until it can be attached to a state of the art unit - which might be in a week or so. 

 

Tasso

 

You like to speak in terms of 'highly optimised', I like to talk infirst principles aroud what is actually optimised.

 

A two-box soluion effectively renders the storage network attached as you know. 

 

There are two things you can alter depending on whether it's local or over a network that can affect SQ via jitter:

  • The power requirements of the drive
  • The volume and periodicity of the data payload (incl interrupts etc)

 

When network attached the first does not exist. You can power most local things independently (M.2 can be a little tricky) but it's recommended as you know. 

 

The volume and periodicity of your data matters a lot. When you have a two-box solution the network only includes anything between the two boxes. Easily replicable with dedicated subnets etc. You also have a short cable, the effects of which are not easily replicable anywhere (I hope you all have your EtherREGENs on very short cables!). If you're sharing SMB or AFP no bueno, you want a NFS share, readonly, with the readahead tweaked up. Any garden variety NAS can do this. Then you're dealing with the size of packets and how often you get them. You're finally dealing with the inherent periodicity in the system dealing with said packets.

 

All is workable with understanding and whilst some of this can get expensive the bulk of it isn't costly.

Posted
On 05/12/2020 at 6:09 AM, rmpfyf said:

The volume and periodicity of your data matters a lot. When you have a two-box solution the network only includes anything between the two boxes. Easily replicable with dedicated subnets etc. You also have a short cable, the effects of which are not easily replicable anywhere (I hope you all have your EtherREGENs on very short cables!). If you're sharing SMB or AFP no bueno, you want a NFS share, readonly, with the readahead tweaked up. Any garden variety NAS can do this. Then you're dealing with the size of packets and how often you get them. You're finally dealing with the inherent periodicity in the system dealing with said packets.

 

All is workable with understanding and whilst some of this can get expensive the bulk of it isn't costly.

 

I have avoided discussing theories because  I saw this issue as being aimed at people who simply wanted  to choose between  a NAS and local SSD storage without further fanfare.  In all the instances I am familiar with, the internal SSD of the server (including  Linear PSU, SOTM/Jcat USB card)  sounded better than when they first sourced the music from their NAS .   Now of course there is a lot that can be done to improve the performance through the NAS  and  I think most people will realise that. However, I think the issue is quite simple for people who were not interested in the  extra investment in money,  time  and complexity to go beyond that initial choice. 

  • Like 1
Posted (edited)
18 minutes ago, TP1 said:

 

I have avoided discussing theories because  I saw this issue as being aimed at people who simply wanted  to choose between  a NAS and local SSD storage without further fanfare.  In all the instances I am familiar with, the internal SSD of the server (including  Linear PSU, SOTM/Jcat USB card)  sounded better than when they first sourced the music from their NAS .   Now of course there is a lot that can be done to improve the performance through the NAS  and  I think most people will realise that. However, I think the issue is quite simple for people who were not interested in the  extra investment in money,  time  and complexity to go beyond that initial choice. 

 

It's even simpler/better than that.    For example, with a local drive plugged into a raspberry pi running something like moOde, not only do you get fast access for mpd to play your music using the moOde interface, the drive also gets shared onto your network for access anywhere, and with a few clicks you also have upnp/dlna functionality.  There's no advantage in putting it on a NAS far away, software/hardware/protocol wise, from the processor and programs doing the work of producing the music. 

 

18 minutes ago, TP1 said:

without further fanfare

 

Yes, it just works, and works well.

Edited by aussievintage
Posted
10 hours ago, TP1 said:

I have avoided discussing theories

 

I'll make it clear. 

 

No theories were mentioned in my last post - all tested fact.

 

If you already have a NAS and are streaming to a Linux box or similar it is generally able to be setup to outperform a local SSD without too much fuss.

 

There are some esoteric cases where local can outperform network, though they're rare.

Posted
19 hours ago, rmpfyf said:

If you already have a NAS and are streaming to a Linux box or similar it is generally able to be setup to outperform a local SSD without too much fuss.

 

There are some esoteric cases where local can outperform network, though they're rare.

 

We are talking about people who have developed a music server and are looking to source music from either a NAS or SSD .  The server would not be a basic add-on to a network but rather one optimised to play audio..  In all cases I can recall with this scenario the network source fell behind the  internal SSD .    

 

Optimising the NAS would improve network performance but  optimising  the server further can improve SSD sourced sound.  Of course there can be exceptions depending   the respective equipment. Ultimately  there is a reason why the very best systems use dedicated for audio usb servers and fully optimised network file servers (over very short connections)  and  not  your garden variety NAS.  

 

 

Posted
9 hours ago, TP1 said:

We are talking about people who have developed a music server and are looking to source music from either a NAS or SSD . 

 

Yes Tasso, I've also spent a lot of time developing music servers, so have others, and have played extensively with NAS or internal drives. I've also developed solutions for very timing-sensitive scientific applications and so tend to look at the problem from applicable first principles. 

 

9 hours ago, TP1 said:

The server would not be a basic add-on to a network but rather one optimised to play audio..  In all cases I can recall with this scenario the network source fell behind the  internal SSD . 

 

 

My experience differs quite significantly. Assuming a network connection is needed (and this is a big assumption - whether for control, streaming, whatever - if not, get WTFplay or similar and stuff it all it RAM) then no, better results were had running data through the network connection for a few reasons:

  • Whilst both can cache a tune to limit reads during playback, with a local drive it's impossible to maintain complete idle. Accordingly there's interrupt and power system variances.
  • Local reads have no avenue to tune towards periodicity; resulting jitter is random.
  • There's additional EMI in the case/on the motherboard traces/etc. 

 

If you put it elsewhere on the network, the power system is independent, packet timing is easily managed and resulting EMI concerns the Ethernet cable  you're already using. 

 

9 hours ago, TP1 said:

Optimising the NAS would improve network performance but  optimising  the server further can improve SSD sourced sound.  Of course there can be exceptions depending   the respective equipment. 

 

 

There's no basis in fact for this statement. Your streamer reacts to the network. There's a packet waiting, there's an interrupt generated, your CPU needs to do something on the network for the moment and route data accordingly. You want this to happen as little as possible and as regularly as possible. 

 

Network performance is audio performance. It is not mutually exclusive as you suggest. 

 

9 hours ago, TP1 said:

Ultimately  there is a reason why the very best systems use dedicated for audio usb servers and fully optimised network file servers (over very short connections)  and  not  your garden variety NAS.  

 

 

Honestly the best reason I can see is to extract money from willing wallets.

 

Whether it's a timing or throughput or power perspective, there is nothing in audiophile requirements that wasn't solved in industry decades ago. Ethernet, even very tuned ethernet, is not new. 

 

I'm a believer in a two-box format, Tasso, though the underlying rationale is simple:

  • A network between two boxes only limits traffic. The NIC on the streamer isn't constantly checking to see if it needs to deal with everything. You can do this with a good switch/router, firewalling or sub netting. 
  • A short cable, preferably optical though other means exist, limits jitter through reduced EMI. However this can also be done with a switch and isolation just upstream of your streamer, being effectively what the etherREGEN is intended to be, though other means exist. (This said, I'm amused at the length of cable many are putting after their etherREGEN... it should be as short as possible.) 

 

Whether the NAS is garden-variety or not matters not a jot. Sorry, that's not how network equipment works. Once the data is on the network it's a packet with some instructions on where it's meant to go and some intrinsic timing. Only the timing changes if you change the power supply to a LPS, use golden hard disks or put EMI absorption everywhere. And you can change headers and timing downstream with secondary devices. A NAS will only ever affect things if it can't keep up, and audio applications are not particularly high-throughput. A garden-variety NAS on 100Mbps Ethernet from 2005 will do let alone a garden-variety NAS from today. 

 

The dudes behind Roon play to a particular audience that needs convenience. If the words 'NFS' are so foreign that you don't want to get to know them then yes, an internal drive is for you. But unless you don't need to be connected to any network then no, network media works just fine in an audiophile context to a streaming PC, and is a better bet than an internal drive. I don't mean to argue with your for the sake of it either. 

 

You don't have a streaming PC. You have a streamer based around an ARM platform and I've no idea how Ethernet is integrated there. Your mileage may differ. What you do have is a very nice reclocker downstream of it all that is absolutely the way anyone should be doing it. 

  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...
To Top