NDX user experience
Posted by: Rich27 on 04 May 2011
I have had my NDX for a week now and apart from a few set up hiccups (mainly my lack of technical ability) the system is sounding better than ever! What a wonderful piece of kit (replaced a sonos/nDAC combo).
I am using the NDX wireless at the moment and this is stated by Naim to be sub-optimal. There is no way that I can run a network cable from my router to the lounge so would it be worth setting up some kind of wireless bridge to the NDX (e.g. Netgear WNCE2001)? At least this would be N not just limited to G (as I believe the NDX is) and would move the nasty wireless bits away from the NDX.
Thoughts appreciated as the only other option would be powerline plugs which I hear introduce all sorts of junk onto the mains.
By the way nStream app is pants, please make this worthy of a £3k piece of kit!
It seems the biggest advantage of using no apple product is being able to be totally indifferent to these long n-serve/n-stream threads!
Allen
I think you're still missing the point–access through the n-Serve app uses no UPnP whatsoever–it accesses a system software service (mezzanine) called Sophia. This interacts a priori with the built-in SQL database via .NET (for those interested why the Desktop Client Installer implores you to use the latest version of the .NET framework).
This layer of access is NOT, however, an absolute guarantee of performance, as anyone who has compared SongBook DigiFi to n-Serve can tell you.
As for n-Stream–UPnP/DLNA control, by its very nature, is "plug-n-play", but updates from the server are not automatically broadcast out to the control points unless they refresh their content listing, and certainly not at a system level (largely for security purposes, among other things).
In the Naim servers, the UPnP server layer interacts with the built-in SQL database, and presents a browse tree directly based on the structure of content within the database itself.
This keeps the demands on the control point down–something has to passively (at the same layer of access as any other control point device) parse and traverse the lists generated at the server end of things, generating its own a posteriori database (possibly using SQlite on iOS-dunno?)....
As for updates–you can cache all of the database at any point in time (with the possibility that it could be wrong at some time in the future due to changes in absolute file location, as explained above), you can prioritize updates based on addition to the database (which means that the UPnP server must be aware of changes from system services at the content-aggregation end of things, e.g., that a CD rip has been added, or a new network share has been added with new material scanned and added to library), you can try to look for new additions manually (!), or, you cache nothing, which slows everything down to a crawl but is accurate at the point of request to the server.
Somewhere in here lies a sensible approach–I'm not sure I'd know what it is.
Good point likesmusic.
It still comes down to the fundamental question, if Linn / Chorus DS, Logitech / Squeezebox, Sonos can deliver the instant browsing capability on a UPnP based system, why can't Naim?
I've seen Sonos systems come to a crawl in the same manner–and these are some of the easiest setups to get working out of the box (in theory). The mesh networking employed (dedicated) can be hung up on some odd but simple configuration no-nos as well as anything else.
Yelling "fire" in a crowded theatre doesn't make the doors any bigger–you need to insure that you have adequate bandwidth (and then some), regardless of network protocol. Too many people here look at the short view ("it works, mostly") than the long one ("we have MORE than enough capacity").
Don't be surprised if you see a Sonos hardware upgrade between now and CEDIA (2011 US).
Might be interesting for one of you N-Stream users who are experiencing slow response times to borrow or buy a second wireless router and see if a dedicated network makes any difference at all. IIRC, it was Simon who pointed out that wireless ethernet is collision based, so its performance is not determinitistic (i.e., you can never be sure how many network packets are competing at one time to be the one that gets through next). My guess is that it will have minimal effect, but who knows...
By the way, is a dedicated network now an official Naim requirement? Is that documented somewhere? Just curious.
When I get home, I think I might just buy another wireless router and dedicate it to the relatively small amount of traffic between my PC music server, NAS and Mac Airbook (used for remote control). Just noticed that products like the D-Link DIR-615 are down to $60 (which seems like a pretty cheap tweak to me).
Hook
Ask your retailer–I've noticed that the NETGEAR WNDR3700 is now down to $129 US (about $50 less) and is a great choice for dual-band routing / bridging / access.
In the US, yes–we recommend a dedicated network–as does NetStreams for the DigiLinX / NaimNet side of things. It's simply good common sense, not marketing spiel or opportunism.
Relative addressing refers to a relational music database storage model where, rather than feeding out files from an explicit, absolute filesystem location within the directory structure, the database uses pointers that reference the file relative to its present location, with its metadata, permitting the movement of the file from one location to another within the storage pool without crashing or having to re-scan the database. Some UPnP control points will cache absolute file resource locations, which causes problems if music is moved between updates–so you have to be aware of the possibility of tradeoffs.
More cod-technical gobbledegook.
I've worked with relational databases for many years; relative addressing has got absolutely nothing to do with them. Indeed one of the criterial features of relational databases is that they do not use pointers to reference data; such things are anathema to relational database design.
A true relational database will involve something like SQL - Logitech's squeezecenter uses one such for example, and with some knowledge of SQL and Erlands plug-in you can query it directly if you so wish.
Not sure where you learned about relational databases, but pointers are one of the components key to the functionality of B-trees in the first place.
On stable versions, the speed of Squeezebox Server is fundamentally down to an excellent PERL scripting implementation which interacts with the database (MySQL or SQlite–your, choice on newest Server builds from 7.6.x above)–the basic underpinnings of the software are pretty straightforward, albeit intentionally limited.
"Gobbledegook" is a stick or a stone. Spend some time with the problem first, instead of guessing, in the interest of credibility.
There are quite a few technically astute people lurking here, and it's not difficult to anecdotally match the Forum username with the retail client's name, as a few dealers have pointed out to me lately–same with the other Forums. The pawprint is the same.
Lastly–those of us who do use our real name stand behind our posts–it would be good that more people were willing to do the same, in the interest of taking responsibility for their opinions.
Bit grumpy there Big D
Understand perfectly why N-Serve is quicker (Sophia - lovely name for a bit of software) - despite running on .net a truly frankensoft creation whose latest bug filled rolling update trashed our email servers last month. In the same way that SB and DAAP will be quicker since they don't rely on UPnP protocols.
Agree that the speed of the UPnP server will largely dictate the speed of N-Stream - that is obvious and I'm now reaping the benefits of the speed increase on a fast Ipad2. I think the relational database implementation is a bit of a red herring as you rightly point out that Squeezebox is fast because it has been very efficiently coded from the outset and a lot of thought has gone into how the MySQl database is queried. I think in retrospect that you are right in saying library tree caching is not going to improve things but perhaps the answer may be a different approach to building the app in the first place. Since I don't know what approach Naim have taken in building their app it would be pointless to speculate.
N-Stream remains the best (small field / pretty ropey competition) solution currently available for Naim. I nolonger find it slow as my server is fast, it doesn't crash anylonger and the last update only took 2 months to appear.
So if UPnP is flawed...
So consider will Naim feel obliged to embrace DAAP when they take a spin with Airplay? Since DAAP is nolonger an Apple only protocol and can be licenced or hacked would it make a better alternative to UPnP in the future and might free up Naim from building CPs in the first place. the view that UPnP is open source is a bit misleading since most vendors rely on extensions and the UPnP/DLNA commitee seem incapable of ensuring true compatability across media - anyone tried DLNA with a Sony Bravia?
Squeezebox or Vortexbox on a UnitiServe - licenced and tweaked - unthinkable? or just a practical cost effective solution to avoid reinventing the wheel every ten minutes.
Does a proprietory (magic rip?) server solution in a lovely produce the best sound or is it an expensive indulgence? Is the idea of music aggregation across a network via music shares brilliant or simply a complication that makes the database's job or caching the tree (and N-stream's role) harder. Surely the whole point of a server is that you store your stuff in one place and back it up regularly?
I suggest that it seems those people running a UnitIServe / HDX UPnP are by coincidence the very people least happy with the performance of N-Stream.
How do those people who run other server configurations find N-Stream?
Or is UPnP still worth following in the hope that efficient database management will one day produce a super app?
Answers on a postcard ....
Oh and David .... (if that is your real name) .... I've always been called Tog
Tog
Not sure where you learned about relational databases, but pointers are one of the components key to the functionality of B-trees in the first place.
Not sure where you learned anything - a b-tree is a completely different entity from a relational database. I think I implemented my first b-tree at university in 1969; the technique had been around for a few years and certainly pre-dated relational database theory.
I learned about relational models of data from the work of Codd (hence my 'cod-technical' dig - you must've missed the pun), and I learned about relational databases from many years working with Oracle for a living.
Very little software is un-improvable.
Sorry, I was thinking that to be a self-referenced inside joke such as "codpiece", aptly chosen.
Codd's rules (for relational databases) are yesterday's news (and were decades ago)–nearly every (non-)commercial relational database fails at least two or more of these–including Oracle's solution, as well as MS SQL and SQlite. You're smart enough to know the application for B-trees at a low level within an RDMS....
Perhaps we shall christen you "likesmusicprefersdatabases"?
More importantly–in a music library, where one expects to have some sort of control over the seek and queue process for a set of music files, empty fields for Artist or Album won't get us anywhere–nor will this work for music storage (filesystems need directory names). So list processing is pretty important–especially for large libraries, bandwidth needs to be properly managed.
The amounts of data involved here are so small that response times should be near instantaneous if the rest of the system is decently implemented. Caching would seem to be an obvious technique, especially as the data is relatively static - iPeng does a decent enough job caching Squeezecenter for example. And I notice that squeezecenter itself can access any UPnP server - just tried it with Asset UPnP across a wireless network and response times are excellent.
Careful with that axe–Squeezebox Server does NOT consistently update its library.
I am nearly always having to do a clear and re-scan for it to pick up new entries, artwork, etc. which should be consistent and automatic, even on an iTunes library with a limited number of albums.
David,
Within Squeezebox Server I think you can set it to scan at any given time frame witin a 24 hour period which all though is not an "automatic" feature, it is useful to set it during "sleep" time.
I also demoed the NDX as a replacement for my Sonos as I wanted a complete Naim line up. The fact that NDX is a full width box didn't hurt either. Sadly, the user experience going from Sonos to N-stream (on an iPad) was a show stopper for me. I was missing playlists, actively adding to and editing the current queue, browsing library in the same window, fast forward, display track length etc.
I am a bit surprised by a previous post that Naim ask what users want improved in N-Stream, to me that shows they don't know much about what is out there (unless ofc it was a question of what to do first). Buy a Sonos ZP90 and test their iPad interface would be my recommendation.
I ended up buying the US because N-Serve is a lot closer to a decent user interface, although not quite at Sonos level yet.
Careful with that axe–Squeezebox Server does NOT consistently update its library.
I am nearly always having to do a clear and re-scan for it to pick up new entries, artwork, etc. which should be consistent and automatic, even on an iTunes library with a limited number of albums.
Sure, squeezecenter has problems - one of which is mismanaging it's artwork cache such that complete re-scans are occasionally necessary. But, as Dustysox points out, this can be done automatically so isn't a big deal. In the past Logitech have had many more performance issues, but they seem to take most problems on board, acknowledge them and fix them - beta releases and nightly builds are even made available to keen users to help the development process. iTunes apps like iPeng work fast, well and reliably. You seem to create pseudo-technical 'explanations' for every naim problem that actually preclude any improvements!
Naim don't exist in a vacuum and need to produce an app that can compete with the competition in a world where people expect a world class control point to go with the "world class sound!" N-Stream is now much better but the competition is fierce.
Tog
Lastly–those of us who do use our real name stand behind our posts–it would be good that more people were willing to do the same, in the interest of taking responsibility for their opinions.
Dave -
I have thought about this for a while. I came to the conclusion that, even in a relatively tame environment like an audio forum, there are still plenty of nut jobs lurking (you know who you are! ), and I simply feel more comfortable keeping a relatively low profile.
That said, Peter, my Naim dealer, knows what my forum alias is. You, too, know who I am and where I live (scary thought that ). As a compromise, I have set up a google mail account and placed it in my profile. I use it only for private conversations with other Naim forum members. Anyone who emails me and uses their real name, will get the same in return from me.
So, if I have ever said anything on this forum that makes anyone think I am hiding behind an alias, feel free to drop me an email. Unless, of course, you are a nut job.
By the way, and perhaps like Tog, Hook has been my real world nickname since I was a young teen. Yell Hook in a crowded room, and I will turn my head to see who called.
Hook
Like Hook , those who need to know ,know my "real" self.
I continiued to follow advice I was given when first using forums, that was to hide behind a nom de type.
No sinister reasons, but it does seem to upset a few forumites that some of us choose to be slightly aloof.
Short of googling a real name, to get info, what friggin' difference does it make if we use a pseudonym?
Also like Hook , H or BigH have been with me for over half a century, so to whom it offends , TOUGH.
Maybe a new thread might be a better place to continue this if needed?
> Lastly–those of us who do use our real name stand behind our posts–it would be good that more people were willing to do the same, in the interest of taking responsibility for their opinions.
That is an unfair comment if you were Christened as Vomitnose Boneidle would you not jump at the opportunity to use a more socially acceptable monicker.
All the best, Guy
ROTF........etc
Personally, I would like to keep the pressure on.
If anyone's listening
I'm listening! Smoke, mirrors and red-herrings abound! Not to mention collisions, flawed headers, and dodgy pointers.
It would be ironic if N-Stream was slowest with the UnitiServe. At the moment the speed of the server and the IOS device seems just as influential on performance as the app code.
Speed is not the issue for me - capability is an issue - playlists anyone?
However, like Allen I'm not convinced N-Stream can't be speeded up and developed a lot quicker than it has been.
Tog
Agreed
Will Naim invest the development costs if they see Airplay/DAAP compliance on the horizon?
Given their current speed of software development .... ?
Tog
I am a little confused. If you have, for example, a Serve AND an NDX what is the point of using nStream? NStream allows control of the preamp volume and source selection right? And the Serve doesn't right? Are there any other functional differences? Or am i just stupid and the nStream is only used for Naim servers that are connected to a Naim renderer.
I also dont understand why there are two apps but that might be a question for another thread. I am guessing I dont get it since I have never used a UPNP renderer like the Qute or NDX. The only UPNP Renderer I have used is my iPhone with PlugPlayer or with J River serving to itself. Cant there be one app that simply show a list of servers and renderers. Then you simply pick one?
Either way, PlugPlayer is just as fast as the Remote App (faster with building the thumbs). I can control the volume on my server or renderer with the same volume slider in the app. I can browse the library and playlists just like remote app. I can create playlists easily etc....
What am i missing here? Just when i think I understand this whole streaming fiasco, I read through a few threads here and feel like my 68 year old father when he tries to do anything with a computer. Maybe another source of my confusion is that my UPNP server is associated with a studly media player. Is it silly for me to think there shouldn't be another way? With J River the UPNP serving is absolutely mindless.
I know, i know........ for unknown reasons (not ONE SINGLE PERSON EVER responded to my comments and questions regarding why the vocal participants here "hate" PlugPlayer), no one like PP. But for me it is absolutely stellar and indispensable. I honestly cant find a single fault with it.
-Patrick
> Lastly–those of us who do use our real name stand behind our posts–it would be good that more people were willing to do the same, in the interest of taking responsibility for their opinions.
That is an unfair comment if you were Christened as Vomitnose Boneidle would you not jump at the opportunity to use a more socially acceptable monicker.
All the best, Guy
My wife's family was a Fawkes, some generations back. I could live with Vono bon Idle though.
The speedup with n-Stream v2.1 made me call my dealer and ask for a "time out" with my NDX demo.
I was sure network performance wasn't the bottleneck so I started to mess around with various different UPnP setups & servers.
Now I'm blown away by the difference they make with NDX performance. It's incredible, really.
The worst setup (so far) being the EyeConnect and the fastest VortexBox & "lowly" PS3 Media Server.
(I've my library as ALAC, so seamless on the fly transcoding to WAV/PCM is a must have for now.)
Anyway, the current setup is probably 20 times faster than the laggy crap I started out with. I still have to find a perfect combination of transcoding and streaming options, but this sure feels like a totally different machine now.
I'm turning my coat on this one. NDX is a beast.
Still, I wonder how less computer savvy customers (or dealers!) will ever get this monster to show it's true colors...
Absolutely - thanks murkku
Similar results with Uniti - almost gave up last year when using N-Stream with EyeConnect - like surfing in treacle - abysmal. Asset UPnP was better, Allegro was slow and Songbook crashed my Mac repeatedly.
You have confirmed my belief that Vortexbox is pretty swift which given that it uses a low overhead Fedora OS and MiniDLNA (developed from the ReadyNAS UPnP server) this should mean that ReadyNAS users will see an improvement as well.
I still suspect from the views expressed so far that David Dever is right and the speed of the server is paramount with N-Stream and that UnitiServe may be the cause of some of the slowness found by others.
Ipad 2 has a pretty dramatic impact as well.
So what UPnP servers are people using and how fast is their N-Stream experience ?
Tog