Minim server + Synology NAS kills custom dedicated pc on music sound quality

[sleask]

I just have to share this story with everyone on this forum as well, my original story is over on PS audio’s forums at http://www.psaudio.com/forum/#/discussio...nful-story
if you don’t want to read the long version, here is the short one;
I spent a lot of time and money testing and building a dedicated music pc, I have always wanted to try minim server and I had an extra synology nas, the ds213+ with 1 wd 3tb red so here was my chance, reset the synology, reinstalled dsm (complete steps at bottom of post),installed minim server, the result was quite surprising, it didn’t sound as the pc but darn close, I thought the pc has ssd’s in it so let’s try one in the synology. OMG it killed the pc, just to make sure I wasn’t totally insane, I invited a fellow audio friend over to confirm the differences, he was blown away.
So if you’re using your NAS as music server only, feeding a network renderer, I highly recommend you try this!
Setup:
Remove hhd installed one ssd, complete unit reset(holding down reset button twice), reinstall DSM, check for updates, set up ssd-synology hybrid raid(no data protection), installed synology’s media server just to get all the systems default shared folders set up, added user and set permissions for the folders, stopped the media server, installed java, installed minim server, installed minim watch on pc, open minim watch and set transcode to “wav24”. Transfer music files, stop windows and mac file services, enjoy!
I noticed some people were having cpu usage problems with 24/192 flac
The DS213+ has the dual core PPb 1.067 GHz chip and here are the systems cpu usage specs transcoding FLAC to wav24 with this set up,
16/44 – 3-4 %
24/96 – 7-12 %
24/192 – 11-17 %

Wattage used by each unit
DS213+ 9w
PC 76 -81+w

Simon, thank you so much for your work and software, you real have taken my system to a new level
Cheers
Steve

[simoncn]

(13-11-2013 01:59)sleask Wrote:  I just have to share this story with everyone on this forum as well, my original story is over on PS audio’s forums at http://www.psaudio.com/forum/#/discussio...nful-story
if you don’t want to read the long version, here is the short one;
I spent a lot of time and money testing and building a dedicated music pc, I have always wanted to try minim server and I had an extra synology nas, the ds213+ with 1 wd 3tb red so here was my chance, reset the synology, reinstalled dsm (complete steps at bottom of post),installed minim server, the result was quite surprising, it didn’t sound as the pc but darn close, I thought the pc has ssd’s in it so let’s try one in the synology. OMG it killed the pc, just to make sure I wasn’t totally insane, I invited a fellow audio friend over to confirm the differences, he was blown away.
So if you’re using your NAS as music server only, feeding a network renderer, I highly recommend you try this!
Setup:
Remove hhd installed one ssd, complete unit reset(holding down reset button twice), reinstall DSM, check for updates, set up ssd-synology hybrid raid(no data protection), installed synology’s media server just to get all the systems default shared folders set up, added user and set permissions for the folders, stopped the media server, installed java, installed minim server, installed minim watch on pc, open minim watch and set transcode to “wav24”. Transfer music files, stop windows and mac file services, enjoy!
I noticed some people were having cpu usage problems with 24/192 flac
The DS213+ has the dual core PPb 1.067 GHz chip and here are the systems cpu usage specs transcoding FLAC to wav24 with this set up,
16/44 – 3-4 %
24/96 – 7-12 %
24/192 – 11-17 %

Wattage used by each unit
DS213+ 9w
PC 76 -81+w

Simon, thank you so much for your work and software, you real have taken my system to a new level
Cheers
Steve

Thanks for sharing this!

I'm impressed that the DS213+ can do FLAC to wav24 transcoding with only 17% CPU usage as well as consuming only 9W. The PowerPC processor used in this model is an unusual choice (Intel and ARM are much more common). From this evidence, it has a lot going for it.

[Linn_agnostik]

Thanks for sharing..

my experience is more nuanced - I use minimserver on my Synology DS-412+ (dual core atom, 2,13ghz), feeding my linn renderer (Akurate DS/1) and I've experimented with transcoding recently.
Tried both wav and wav24 and thought I lost a bit of warmth and grit in (predominantly) voices, both male and female. Turned off transcoding and lo and behold, the warmth and organic nature returned.
No major differences, not night and day in any sense, but clearly discernible.
I have no ssd in my system and unfortunately no way to test that, as it is fully loaded

Wav24 transcoding from flac on 192/24 was <5% cpu utilisation

[simoncn]

(14-11-2013 22:01)Linn_agnostik Wrote:  Thanks for sharing..

my experience is more nuanced - I use minimserver on my Synology DS-412+ (dual core atom, 2,13ghz), feeding my linn renderer (Akurate DS/1) and I've experimented with transcoding recently.
Tried both wav and wav24 and thought I lost a bit of warmth and grit in (predominantly) voices, both male and female. Turned off transcoding and lo and behold, the warmth and organic nature returned.
No major differences, not night and day in any sense, but clearly discernible.
I have no ssd in my system and unfortunately no way to test that, as it is fully loaded

Wav24 transcoding from flac on 192/24 was <5% cpu utilisation

It is interesting that you noticed a difference and preferred the FLAC sound to the WAV sound. For me, the WAV sound is preferable, but I have no way of knowing which of these is closer to the intentions of the engineer who created the recording.

[magister]

(13-11-2013 07:54)simoncn Wrote:  I'm impressed that the DS213+ can do FLAC to wav24 transcoding with only 17% CPU usage as well as consuming only 9W.

Just to give some comparison for those who are contemplating a Synology purchase: as I type this, my DS213 (not the "+" version) is playing a 24/176 file, being transcoded from FLAC to WAV24 by MinimServer. According to the Resource Monitor built into DS 4.2, CPU usage ranges between 20-30%, while the memory usage just sits there at 15%. So I would say there is no need to spend money on any higher level Synology machine, although the results from the 213+ are certainly impressive.

[DavidHB]

What is interests me is that Steve and his friend got what they felt were significant audio improvements when the drive in the Synology was changed to an SSD. I'm at a loss to understand how this could be, unless, perhaps, the SSD firmware allows it to work better with the bus on the Synology, and thus feed a 'better' audio stream to its Ethernet port.

The other thought (which was implicit in some of the responses to Steve's report in the forum thread to which he linked) is that the results may have something to do with the architecture of his system. The PS Audio DAC with add-in network 'bridge' represents a very different approach from the Linn DS/DSM models which I think at least three of the responders on this thread are using; some posters on the other thread were even suggesting that there is an audible difference between different versions of the bridge firmware.

What is not in doubt (and Simon has been saying this all along) is that MinimServer, even when doing WAV24 transcoding, does not require massive hardware resources to run well - one of the many feathers in its cap.

David

[simoncn]

(17-11-2013 18:34)DavidHB Wrote:  What is interests me is that Steve and his friend got what they felt were significant audio improvements when the drive in the Synology was changed to an SSD. I'm at a loss to understand how this could be, unless, perhaps, the SSD firmware allows it to work better with the bus on the Synology, and thus feed a 'better' audio stream to its Ethernet port.

The other thought (which was implicit in some of the responses to Steve's report in the forum thread to which he linked) is that the results may have something to do with the architecture of his system. The PS Audio DAC with add-in network 'bridge' represents a very different approach from the Linn DS/DSM models which I think at least three of the responders on this thread are using; some posters on the other thread were even suggesting that there is an audible difference between different versions of the bridge firmware.

What is not in doubt (and Simon has been saying this all along) is that MinimServer, even when doing WAV24 transcoding, does not require massive hardware resources to run well - one of the many feathers in its cap.

David

The audio stream data is taken from the contents of the file with wav24 transcoding applied. Nothing in the disk or bus would change this data.

However, it's possible that the TCP/IP message sequence (buffering, retries, etc.) could be affected by disk and bus issues. Perhaps this is what you meant by a "better" audio stream. For most renderers, this shouldn't make any difference because the renderer will put the network data into an internal buffer before feeding it to the DAC. Perhaps the PS Audio DAC/Bridge is more sensitive to TCP-level differences than other renderers.

[sleask]

DavidHb, I don't know why either and i don't think we will find out, so i'll take your reason, lol, so i can drop this task of "why" and just start enjoying music again, as posted in the ps forum i did "temporarily" purchase the DS412+ for further testing, please forgive me but i'm just going to copy and paste, i'm exhausted after all the testing and thinking.
. As far as performance goes you don’t need the higher end models, I did compare the 213+ with the 412+, I did prefer the 213+ buy a margin but it could be a “break in” factor, the 412+ was newer. (I did run the unit hard for two hours before running the tests).
As far as raid goes, I have read many times not to run the ssd’s in raid, double the write times = half the life of the drive. Also, my buddy at the computer store said if you’re not running raid in the nas you can install any dive you want, the compatibility is only for the raid function. I have about 1000 albums all in flac, this is about 439gig, but, I’m sitting here thinking before I go buy a large ssd, do I really need the whole collection for serious listening? Not at all, I’m sure I won’t listen to 80% of it. BUT what if I rerip everything to wav, oh the pain, the biggest factor is I’m male and don’t think very rational so I will probably go for the 1 tb.

My conclusion is ,
Does the nas sound better than the pc, absolutely
Is the nas a cheaper route than a high end pc, absolutely
Do you need the top of the line model with all the power, “for audio”, absolutely not.

there is one other thread going at ps on this at http://www.psaudio.com/forum/#/discussio...s713-drive

just a note on the system resources, if you go into the process tab,and watch you will see that the process monitor itself is peaking upto 2.2% at times, i'm gessing it's java you want to watch, so the minim server is even lower than reported(the higher peaks)
Again great job Simon!
Linn_agnostik, thank you for the suggestion, i will try all the settings

steve

[DavidHB]

(17-11-2013 18:52)simoncn Wrote:  ... it's possible that the TCP/IP message sequence (buffering, retries, etc.) could be affected by disk and bus issues. Perhaps this is what you meant by a "better" audio stream.

Yes, that is what I meant, and why "better" was in quotes. We all know that tiny timing differences (jitter or whatever) can affect sound quality. Whether this is what is happening here is pure speculation, of course.

Quote:For most renderers, this shouldn't make any difference because the renderer will put the network data into an internal buffer before feeding it to the DAC. Perhaps the PS Audio DAC/Bridge is more sensitive to TCP-level differences than other renderers.

Well, possibly. But we don't know that other renderers (including ours) would not be similarly affected. We do know that Linn, to take but one example, try to eliminate timing differences by locking the stream to a clock at the DAC end (while presumably also using a buffer), and they would not do this unless timing issues in the stream (to which TCP differences no doubt contribute) affected sound quality.

All that said, and though speculation like this is fun, I have to go along with Steve's preference for "just enjoying the music", even though he does seem to be in danger of backsliding ...

David

[DavidHB]

(17-11-2013 19:09)sleask Wrote:  DavidHb, I don't know why either and i don't think we will find out, so i'll take your reason, lol, so i can drop this task of "why" and just start enjoying music again.

You're 100% right, but that pesky "why" just keeps on intruding, for you and me both, it seems . It's why I spent much of yesterday rewiring my speaker connections (for reliability and maintainability, rather than sound quality) ...

Quote:I have about 1000 albums all in flac, this is about 439gig, but, I’m sitting here thinking before I go buy a large ssd, do I really need the whole collection for serious listening? Not at all, I’m sure I won’t listen to 80% of it. BUT what if I rerip everything to wav, oh the pain, the biggest factor is I’m male and don’t think very rational so I will probably go for the 1 tb.

Don't forget that you can have more than one instance of MinimServer on the network (though not - yet, at least - on the same device), so you could start by putting your 50 or 100 most played albums on a smallish SSD, and keep the main collection on its spinning drive. Following your own precept, you ought to be listening rather than ripping - unless you can do both together, of course .

Quote: My conclusion is ,
Does the nas sound better than the pc, absolutely
Is the nas a cheaper route than a high end pc, absolutely
Do you need the top of the line model with all the power, “for audio”, absolutely not.

And, by any standards, those are interesting and useful conclusions, and I'm grateful that you started the thread.

David