Euphony OS w/Stylus player setup and issues thread

[Lukasluis]

I tried to scan the thread but I can’t seem to find how to load convolution filters to HQPlayerD. Can you point me to the guide please? Is this even possible with Euphony?

[Lukasluis]

1 hour ago, AnotherSpin said:

 

I am in a process of trying ramroot with NAA endpoint Euphony. First impression is very good! With this I want to confirm a definite change with enabling ramroot.

Is this feature available on the Trial license. I can't seem to find the option for Ramroot in the Expert settings. 

[Lukasluis]

42 minutes ago, AnotherSpin said:

 

Please refer to @austinpop post above (#439)

Thanks for the pointing it out. I shall open a support ticket.

[Lukasluis]

On 8/17/2019 at 6:22 PM, ASRMichael said:

Lukasluis, email Euphony they will send you trial version. Note it's not a finished version, but worth testing and feeding back. I;m in the process the now. 

Thanks I am testing it now. 

[Lukasluis]

I've been playing around different combinations of cpu freq, turbo and hyperthreading for the past 2 days. I found that for my system a fixed frequency of 1.2 Ghz, turbo on is the best (I can't set it lover than 2.1Ghz if it's off). My server is running Roon/Euphony, Intel NUC8I7 in Akasa Turing case, root in RAM. Roon is processing minimal PEQ taps to correct for room modes and 1.2ghz is responsive enough for my ~100K tracks library and Tidal integration. I found that the higher I set the frequency the bigger it sounds but more accurate timber of instruments and voices are more natural at the lower frequencies. I found that if the room modes are uncorrected, the music are slow, boring or dragging with lower cpu frequencies. 

 

I would like to experiment on the effect of CPU isolation but I don't know the name of the applications/processes when running Roon/Euphony. 

[Lukasluis]

4 hours ago, Lukasluis said:

I've been playing around different combinations of cpu freq, turbo and hyperthreading for the past 2 days. I found that for my system a fixed frequency of 1.2 Ghz, turbo on is the best (I can't set it lover than 2.1Ghz if it's off). My server is running Roon/Euphony, Intel NUC8I7 in Akasa Turing case, root in RAM. Roon is processing minimal PEQ taps to correct for room modes and 1.2ghz is responsive enough for my ~100K tracks library and Tidal integration. I found that the higher I set the frequency the bigger it sounds but more accurate timber of instruments and voices are more natural at the lower frequencies. I found that if the room modes are uncorrected, the music are slow, boring or dragging with lower cpu frequencies. 

 

I would like to experiment on the effect of CPU isolation but I don't know the name of the applications/processes when running Roon/Euphony. 

Sorry, I'm running Roon/Stylus not Roon/Euphony. 

[Lukasluis]

2 hours ago, TheAttorney said:

Just hit the Core Isolation Apply button. This will display all the relevent processes. As your default position is an empty Core Isolation field, hitting the Apply button will display current values without changing them.

Thanks I see bunch of tasks including 3 Roons, Stylus, nfm and gstp? I plan to isolate only stylus, what is gstp and nfm for? 

[Lukasluis]

42 minutes ago, Lukasluis said:

Thanks I see bunch of tasks including 3 Roons, Stylus, nfm and gstp? I plan to isolate only stylus, what is gstp and nfm for? 

Found that gstp is the player part of Stylus. I have now isolated core 6&7 for it. The result is the a notch in transparency and the presentation became more immediate.

[Lukasluis]

I have the same issue. I also could not restore my backup when in Ramroot. Restore keeps failing during the middle of the restore with Backup not available error. 

[Lukasluis]

25 minutes ago, FelipeRolim said:

I downloaded the latest version of Euphony and restarted the experiments. I set it up as I had done before and added the suggestions colleagues left on the topic. Improved, but aspects that do not relate to what I wrote that Windows is superior. That is, it continued to play incorrectly from the timbre point of view, without the same dynamics and articulation of Windows and with a remarkable poverty in textures and harmonics. With the monitor connected to the computers, I tried everything without success. It was the last time I tried a Linux.

I too find Euphony Stylus' timing and dynamics to be off in my system thus I use Euphony RoonCore (NUC directly connected to DAC). Can you try and let us know if switching to Roon makes a noticeable improvement in timing and dynamics.

[Lukasluis]

5 hours ago, lwr said:

 

I too was surprised how much the switch to Apacer RAM further reduced the noise floor, and further improved clarity and soundstage.  Highly recommended     

 

5 hours ago, lwr said:

 

I too was surprised how much the switch to Apacer RAM further reduced the noise floor, and further improved clarity and soundstage.  Highly recommended     

The NUC does not support ECC RAM however reducing from 16GB to 8GB (removed 1 8GB-stick) resulted in more smooth presentation. I'm using Kingston HyperX CL14 RAM, RAMROOT-enabled.

[Lukasluis]

20 minutes ago, Chopin75 said:

I use the AMD/Ryzen CPU and MB which supports the ECC.  I am actually going the opposite direction to need more RAM. I have found that large files suffer from SQ due to not having enough RAM. Though using Apacer seems to have resolved this issue. nonetheless 156 GB is not adequate to play DSD512 and in general I have huge files due to listening to classical music with tracks lasting up to 30 min! So DSD256 and DXD takes up lots of ram such that I am having issues when running in Ramroot + 100% buffer on Euphony OS/stylus. I am likely to do 32 GB RAM in future, which if I use Apacer ECC, I don't think will make SQ worse. 

Supposedly using all 4 rails of the RAM would be better than occupying 2 rails with same total RAM. 

So. I will do 4x 8 GB (already using 2x 8 GB) = 32 GB

2x 16 GB would be cheaper but may not work as well, though it allows 4x 16 GB in future!! Such large RAM might indeed affect SQ as it may perhaps strain the LPS, not sure..

And inject more noise. I'm using a more than adequate PSU.

[Lukasluis]

There is a software update for Roon. Do I need to disable Ramroot first before installing the update?

[Lukasluis]

7 hours ago, auricgoldfinger said:

 

Disable ramroot, reboot, and then install the Roon update.  You'll need to re-enable ramroot and reboot after the Roon update is complete.

Thanks.

[Lukasluis]

6 hours ago, jcn3 said:

 

you can also do the update within roon and then do a save system in AL when done.  that's how i did it.

What do you mean by AL?

[Lukasluis]

18 minutes ago, jcn3 said:

 

I'm an idiot - thought I was in the Audiolinux thread. Please ignore my suggestion as I know nothing about Euphony!

No worries

[Lukasluis]

8 minutes ago, austinpop said:

Indeed, this is what drove the idea. You’ll recall my earlier experiments where I’d prime a queue of tracks, hit play and then pause for the background buffering to happen. This only moved music from its original location to the E_CACHE directory in the /data partition, which is still on the Optane SSD even in RAMROOT.

 

With this new feature, files are actually copied into RAMdisk (technically /tmp I believe), so once buffered this way, you can observe that there is almost no disk or network I/O.

 

I hope people are using the disk and network monitors, by clicking on the word Registered and the build number respectively at the bottom right of the UI?

Yup, even with 100% buffer before play and Ramroot-Enabled, I still see quite a bit of Network and Disk activity around the time when the next track starts. The duration of activity varies on the size of the file being loaded and whether the file is fetched from the NAS or from the local temp drive. I use temp drive for faster loading of tracks. 

[Lukasluis]

38 minutes ago, austinpop said:

Bingo. That’s what this new feature is solving. It’s extending the idea of “Buffer before play=100%” to a set of tracks. It’s a nice SQ lift.

Thanks for working with the Euphony developers in getting these SQ improvements included in the next release. 

[Lukasluis]

19 minutes ago, austinpop said:

 

All credit goes to Željko - he came up with this one himself!

I hope this feature will be extended to Tidal tracks. 

[Lukasluis]

27 minutes ago, davide256 said:

Possible, using Crucial memory in the NUC. Would like to buy Apacer but all I see advertised for NUC are the 16gb sticks of non-ECC, not sure if that would

make any difference.

I changed my Crucial rams in my NUC 8i7beh into these HyperX HX424S14IB2K2/16 Impact... https://www.amazon.com/dp/B0722Q3243?ref=ppx_pop_mob_ap_share

 I also turned off all BIOS devices that I don't use including all USB ports except the ones at the back. I am running my NUC fanless using Akasa case. I also removed all usb enhancers/cleaners/reclockers. No SATA/SSD disks installed nor USB-attached HDDs. These changes make quite a big difference in terms of less grunge (more smooth) sound. I also found that I like the back bottom USB port the most. 

[Lukasluis]

Has anyone compared the sound signatures of Stylus and Roon 1.7 in a single-box  Euphony machine? I find that I like Stylus for Jazz and Classical music and prefer Roon for Rock. I noticed an improvement with Roon after upgrading to 1.7

[Lukasluis]

27 minutes ago, jcn3 said:

 

i just wanted to loop back -- based on your testing, using roon rock on my nuc8i5 server, i tried using AL with squeezelite on my endpoint (i don't have euphony).  i found that squeezelite sounded much better than roon bridge -- roon bridge seemed overly muted/reserved.  squeezelite provided the life and energy that i'm looking for.

 

so with all of the testing and comparisons with AL and euphony, i wonder if it was simply RAAT that creating the difference in sound.

I have the same config for my headphone rig, squeezelite is a little bit better/clearer for music up to 24/96. Roon endpoint is much better for higher res and DSD format. It looks like downsampling is more detrimental to SQ than buffering in memory?

[Lukasluis]

1 hour ago, jcn3 said:

 

i actually didn't do my comparison using higher res files -- i was using 16/44 and 24/96 so there wasn't any up or down sampling.  

 

i don't understand why the higher bit rate/depth files sounded better with roon bridge.  with squeezelite, are you saying you were getting too much clarity (making it sound brittle, maybe?) with higher res files?

I find the opposite, downsampling DSD tracks to Squeezelite 24/96 sounds dead/uninteresting as compared to playing them natively to Roon endpoint.

[Lukasluis]

17 minutes ago, bobfa said:

Over time I have found that running either Stylus stand alone or Stylus EP has sounded better than running RAAT.  For both of my systems the difference has been clear in favor.  As of late I have been testing a different endpoint MoOde OS. 6.4  on an Allo USBridge Sig.  I am finding greatly improved sound quality, even my wife is commenting on it and she never does that.

 

When I run a Roon Server on either ROCK or Euphony to the Squeezelite renderer there is a  Sound Quality improvement. The system is unstable and music does not start right every time I make a change. 

 

I am mostly using Audirvana+ to UPnP .   I hope to test Euphony to UPnP when it gets further along.

 

 

 

 

Eagerly-awaiting for your testing. Any idea when the new version of Euphony comes out?

[Lukasluis]

10 minutes ago, Chopin75 said:

OK, please clarify your question. Why is it surprising downsampling  is worse? Isn't that expected. In general hi Res sounds better. However, upsampling a low res file may not improve things, which can depend on the HW, SW and the quality of the recording. 

I have a bunch of awful DSD256 (like the one I am listening now, which I end up using to test how great a system can be in minimizing the poor quality). A great recording in 16 bit can sound much better than an awful DSD256.  

If you too find that downsampling sounds worse than native then we agree. In my use case, where I have multi-format music collection I find that switching my endpoint’s config every time the music format changes is not sane thus I found the compromise in Roon endpoint. DSD vs 16 bit is irrelevant to me as I only retain one copy of each album in my collection. Yes, I deleted a bunch of crap DSD albums too.