HQPlayer Linux Desktop and HQplayer embedded

[Miska]

On 6/16/2021 at 4:45 PM, Jacky820 said:

Thanks for your quick reply.

Here are the screen cap for the CPU loading. 

So far the CPU 0 and 7 is 94.7% and the rest  is 9.9%  t 12.6%!

 

Is this from HQPlayer Embedded 4.24.1? It should still work at 94.7%, but if those trip over to 100% at some point you'll have a glitch. On my Ubuntu Server system with AMD optimized build with 5800X CPU (and DDR4-3600 CL16 RAM) I get the two cores hanging around 65% load.

 

Have you checked you have correct memory profile active in BIOS settings?

 

[meurer]

I've been trying to package HQPlayer for NixOS (nixos.org) but had some questions; @Miskawhat's the best way to reach you for such questions? I tried email but I'm not sure if your spam filter got me.

 

Questions:

1. Is there a way to make hqplayerd use /var/lib/hqplayer instead of /var/hqplayer? That way it can more accurately follow the FHS and avoid surprising users. With Roon I can do this by setting the ROON_DATAROOT env var.

 

2. What permission groups, other than `audio` does the hqplayer user need?

 

Thanks!

[Jacky820]

On 6/18/2021 at 6:15 AM, Miska said:

 

Is this from HQPlayer Embedded 4.24.1? It should still work at 94.7%, but if those trip over to 100% at some point you'll have a glitch. On my Ubuntu Server system with AMD optimized build with 5800X CPU (and DDR4-3600 CL16 RAM) I get the two cores hanging around 65% load.

 

Have you checked you have correct memory profile active in BIOS settings?

 

@Miska thanks for your reply.

just do the quick check and rest the bios with no OC. the cpu loading down to 84.9%

[lpost]

I'm curious to understand the difference(s) between DSD5EC and ASDM5EC.

 

The descriptions only differ with the word 'rate'. I think I prefer ASDM5EC over DSD5EC but they are very close. Both with 44.1k PCM content.

 

DSD5EC

Rate adaptive fifth order one-bit delta-sigma modulator with extended compensation.

 

ASDM5EC

Adaptive fifth order one-bit delta-sigma modulator with extended compensation.

[Miska]

6 hours ago, lpost said:

I'm curious to understand the difference(s) between DSD5EC and ASDM5EC.

 

The descriptions only differ with the word 'rate'. I think I prefer ASDM5EC over DSD5EC but they are very close. Both with 44.1k PCM content.

 

DSD5EC

Rate adaptive fifth order one-bit delta-sigma modulator with extended compensation.

 

ASDM5EC

Adaptive fifth order one-bit delta-sigma modulator with extended compensation.

 

DSD5EC adapts it's parameters based on output rate. ASDM modulators are more advanced and adapt in various ways.

 

 

[lpost]

I'm afraid I'm back to 4.22.3, again. For my setup, 4.23+ introduced a very annoying white noise issue that I've not been able to resolve except by returning to 4.22.3 which does not suffer with the fault.

 

I've found zero consistency of what tracks will fail and no change of options resolves it. I've also run with no upsampling PCM only noise shaping as I do use gain control. The problem still occurs eliminating system load from the equation. It occurs with PCM, DSD or PCM to DSD and no upsampling as mentioned. ~1 in 10 tracks but occasionally every other track but then it may play 20 tracks without issue.

 

The one thing that is consistent, it only occurs at the beginning of a new track. Simply restarting the track resolves the issue every time. 

 

Path is Roon Core Windows, HQPe to NAA. I'm hoping someone else reports this issue. Seems odd I'm the only one affected.

[sledwards]

31 minutes ago, lpost said:

Path is Roon Core Windows, HQPe to NAA. I'm hoping someone else reports this issue. Seems odd I'm the only one affected.

Well, one way to eliminate one variable would be to put your Roon core on a Linux machine if possible. I have no idea why or how that would effect the music path, but at this point, something easy to try. 

[lpost]

25 minutes ago, sledwards said:

Well, one way to eliminate one variable would be to put your Roon core on a Linux machine if possible. I have no idea why or how that would effect the music path, but at this point, something easy to try. 

I could but I use other Windows apps occasionally and frankly the Windows version runs better than ROCK in my 4+ years experience. The source of the issue is whatever changed going from 4.22.3 to 4.23.0 and beyond.

[Jacky820]

Just now, lpost said:

I could but I use other Windows apps occasionally and frankly the Windows version runs better than ROCK in my 4+ years experience. The source of the issue is whatever changed going from 4.22.3 to 4.23.0 and beyond.

I prefer ROCK rather than Windows + Roon!

 

[sledwards]

4 minutes ago, lpost said:

I could but I use other Windows apps occasionally and frankly the Windows version runs better than ROCK in my 4+ years experience. The source of the issue is whatever changed going from 4.22.3 to 4.23.0 and beyond.

Yes, I understand. Can't speak to ROCK, never used it. I assume you have sent a log file to Miska?  Do you have another DAC to try?

[lpost]

11 minutes ago, Jacky820 said:

I prefer ROCK rather than Windows + Roon!

 

ROCK makes many, many thousands of unnecessary DNS requests and updates 'metadata' frequently. I don't see the same with Windows Core. I've gone back and forth several times over the years, both have advantages and disadvantages, like anything. The key with Windows is find a version/patch level that works and cease all Windows updates and changes. I really don't want to manage another Linux machine. Ugh, sounds too much like work.

 

It doesn't make sense to me to change another variable when I know the one variable that introduces the problem. 4.23.0+. I've tested this theory more than a dozen times. 4.22.3 has never exhibited the problem. Changing the filter, bit rate, DAC bits, buffers, modulator, none of them make any difference.

 

I have Roon set to only send the unaltered stream to HQPe. No DSP of any kind, no leveling, nothing. Something is going sideways in the first packet or two decode when the stream hits HQPe.

[sledwards]

6 minutes ago, lpost said:

It doesn't make sense to me to change another variable when I know the one variable that introduces the problem. 4.23.0+. I've tested this theory more than a dozen times. 4.22.3 has never exhibited the problem. Changing the filter, bit rate, DAC bits, buffers, modulator, none of them make any difference.

So how else do you diagnose this type of problem when your'e the only one having it?  Going from 22.3 to 23.0+  is the cause, and something must be unique to your combination of software and/or hardware.  Sounds like you have been through all the HQPe settings. Again, Miska is the only one who knows the nuts and bolts of the code and its interaction with the OS and some set of hardware. BTW, I have found that Ubuntu 20.04 Server is extremely easy to manage. Write one simple script and run it when the system reports there are updates available...done. Much more efficient than freezing W10 version and patches.

[lpost]

My post was simply to describe what I'm experiencing and see if any others have had this occur with their setup. The HQPe logs don't indicate anything wrong or odd, apparently.

 

Jussi made a couple of different builds of 4.23.1x and the problem occurred less frequently but still occurred. 4.24 is the same as the later 4.23.1x versions.

 

If the problem is with Roon feed, ok, but I'm not to the point where I want change/break a working system (running 4.22.3) and not have music while I build, update and learn a new OS. No need to debate Roon host OS preferences.

[lpost]

33 minutes ago, sledwards said:

Yes, I understand. Can't speak to ROCK, never used it. I assume you have sent a log file to Miska?  Do you have another DAC to try?

I have sent logs, nope, no other dac at the moment.

[sledwards]

4 minutes ago, lpost said:

I have sent logs, nope, no other dac at the moment.

Okay, good luck with your quest for a fix. I am sure something will surface to solve your issue.  Meanwhile, enjoy the May, I sure am.

[luisma]

23 hours ago, lpost said:

I'm afraid I'm back to 4.22.3, again. For my setup, 4.23+ introduced a very annoying white noise issue that I've not been able to resolve except by returning to 4.22.3 which does not suffer with the fault.

You are using the May right? which is known for having some "clicks and noises" when playing content initially? wouldn't this be something specific to the May? maybe the difference between 4.22 and 4.23 is in the way the USB gets initialized or the initial stream sent I don't really know but just saying, even though it happens with the HQP version it might not be an HQP issue....

 

[lpost]

2 hours ago, luisma said:

You are using the May right?

 

I do have the May. I built an Unbuntu 20.04 Roon Server build and so far no white noise issues. The noise isn't the usual click or pop like when switching from PCM to DSD, this is straight up white/pink noise with no musical content.  It really blows the flow of music. 

 

I may have been the changes in HQPe 4.23+ and Windows Roon Server combo. Time will tell.

 

In all cases I'm running a NAA connected to the May.

 

I tried spinning up ROCK as it's quick but the db restore didn't see the USB flash drive connected locally. Lame.

[lpost]

I've got to say, RoonServer on Ubuntu is quite a bit less responsive than Windows 10 or ROCK, on the same hardware. I'm running ROCK now despite getting Ubuntu all setup and working fine because of the latency when accessing Roon. ROCK also does not, so far, exhibit the white noise issue with 4.23+. Seems that something changed with HQPeOS that Windows/RoonServer isn't 100% happy with or vice versa.

[Miska]

2 hours ago, lpost said:

I've got to say, RoonServer on Ubuntu is quite a bit less responsive than Windows 10 or ROCK, on the same hardware. I'm running ROCK now despite getting Ubuntu all setup and working fine because of the latency when accessing Roon. ROCK also does not, so far, exhibit the white noise issue with 4.23+. Seems that something changed with HQPeOS that Windows/RoonServer isn't 100% happy with or vice versa.

 

I have RoonServer on Ubuntu Server 20.04 LTS working fine. I don't notice any difference to running Roon on my iMac. I have not noticed any white noise issues with either two.

 

[lpost]

I believe both ROCK and Windows have some further efficiencies with .Net or Mono that have yet to make it to Linux versions. It works fine but is not as quick to respond. The white noise issue appears to only plague Windows.

[sledwards]

6 minutes ago, lpost said:

I believe both ROCK and Windows have some further efficiencies with .Net or Mono that have yet to make it to Linux versions. It works fine but is not as quick to respond. The white noise issue appears to only plague Windows.

Just curious, where does this latency show up in Roon when running Linux?  Operation seems pretty instantaneous with Roon server running on  my Ubuntu 20.04 installation. The only real delays I see are with some of the long filters loading up on the beginning of individual tracks.

[lpost]

Not in sound latency, it's the interface responsiveness of Roon client retrieving data from the server. It's not a huge difference and I would have likely not noticed it if I hadn't experienced quicker. I also spent no time doing any tuning on Ubuntu either but I did a minimal install so there shouldn't be much left to remove. I suspect it's the older framework Roon Labs is still using on the platform that accounts for the few ms lag.

 

https://community.roonlabs.com/t/roon-client-server-best-practice/976/69 - this is old but I believe still applies based on my experience.

[Saffuria]

Hi @Miska,

can you help me understand what is going on?

It's hqpd os 4.24.1 on an i7-4770(old but it does avx2).

 If a set of filter dither/modulator works, will always work.

Sometimes, trying to play a new combination, one thread on the right goes straight to 100% and there will stay. Browser control unresponsive.

Then I have to reboot and change settings

Even if I send "shutdown -r now" this is what I get 

I was trying below PCM setting.

Below SDM setting plays very nice.  

 

 

[Miska]

23 minutes ago, Saffuria said:

Hi @Miska,

can you help me understand what is going on?

It's hqpd os 4.24.1 on an i7-4770(old but it does avx2).

 If a set of filter dither/modulator works, will always work.

Sometimes, trying to play a new combination, one thread on the right goes straight to 100% and there will stay. Browser control unresponsive.

Then I have to reboot and change settings

Even if I send "shutdown -r now" this is what I get 

I was trying below PCM setting.

Below SDM setting plays very nice.  

 

 

 

poly-sinc-long can take hours to initialize without "Auto rate family" for that setting combination. Less so when going from 48k family to 44.1k family. So this is normal. Initialization time can vary from seconds to hours depending on setting combinations.

 

[lpost]

On 6/22/2021 at 11:32 AM, lpost said:

I've got to say, RoonServer on Ubuntu is quite a bit less responsive than Windows 10 or ROCK, on the same hardware. I'm running ROCK now despite getting Ubuntu all setup and working fine because of the latency when accessing Roon. ROCK also does not, so far, exhibit the white noise issue with 4.23+. Seems that something changed with HQPeOS that Windows/RoonServer isn't 100% happy with or vice versa.

In the past ~10 hours of playback, I've had 3 occurrences of the white noise issue running ROCK. Much better than while running Windows but still not normal. I'm back to Ubuntu 20.04 LTS. Hopefully this problem does not occur again...