HQPlayer Embedded Discussion

Audio Gear Talk HQ Player
141

Technically, there’s ALSA loopback feature. And then pulseaudio can also be configured to do similar. But I haven’t got time to try these out extensively. ALSA loopback at least as a small problem of being stuck when only one end is open. USBStreamer is easy since it is pre-configured in Embedded.

I just today remembered I have PrismAudio Lyra in storage and it has clock slaving support. I will be trying it out for Windows input feature since it supports clock slaving from Toslink input. Even more expensive input device than ADI-2 Pro :sweat_smile:

1 Like
142

Can anyone tell me how to uninstall HQPlayer Embedded? I’ve decided to go with the Desktop version, as it’s license is portable between machines. I’ve installed Desktop on the same machine but, when I check over the network, the web interface is still live. I’ve now removed ~/.hqplayer and var/lib/hqplayer and disabled the system daemon - anything else needed?

143

Did you stop the daemon? But if you are on Ubuntu/Debian, then as usual “sudo dpkg --purge hqplayerd” will do. And on Fedora “sudo rpm -e hqplayerd”. Before doing that “sudo systemctl stop hqplayerd” will stop the service.

1 Like
144

Thanks - had forgotten about using dpkg (my *nix is kinda rusty). Sorted now.

145

PC specs and HQPlayer settings are below.

CPU :i9 13900K
Memory:DDR5 16M x2
closed-form-16M ASDM7EC (PCM>)DSD512

With DSD512, there is no problem.
However, with DSD1024, the sound jumps.

Therefore, I am considering installing a graphics board.
Please recommend a graphics board that will play DSD1024 with closed-form-16M and ASDM7ECv2.

Is an RTX3060 not possible?

What are the key specs required for the graphics board?

Thanks.

146

If you use the HQP OS image, you can get 1024 ASDM7EC and poly sinc gauss hires by itself, no gpu. However, I do have 32gb ram. Don’t know if that matters much.

4090 will not get you to ASDM7ECv2 without stutters at least with my build with Kraken Z63 280mm radiator and Meshlicious case.

If you didn’t see my post earlier, I ended up preferring AMSDM7EC 512+fs and sold the 4090. Maybe revisit next gen. but not something I feel the need to chase any further.

147

I can get 1024 ASDM7EC and poly-sinc-gauss-hires by itself, no gpu.
However, I can’t get 1024 ASDM7EC and closed-form-16M by itself, no gpu.

I want to get 1024 ASDM7EC and closed-form-16M, with gpu.

I would like to know the gpu specs.

Thank.

148

RTX4090

  • cuda cores 16384
  • vram 24gb

Speculation:

RTX4090Ti (not available yet)

  • cuda cores 18176
  • vram 24gb
149

Hi probably a dumb question. I bought HQ Player embedded and was going to installed it on a mac mini, is that possible? If so can someone point me to how to do this? I can wipe the OS if needed also.

150

If this is possible at all it probably depends on the model, If it’s M1/M2 I suspect it’s a bust for sure. If it’s an older x86-64 model that you can get Linux running on then you’ve got a fighting chance:

1 Like
151

thank you.

But I would like to know the recommended minimum gpu specs.

152

Minimum specs will not help in the decision if it boils down to the top 5 cards available right now…

Just a quick amazon search I’ve come up with this…

  • 4080 9728, 16gb ram $1200-1700

  • 3080ti. 10240 cores, 12gb ram $1200-1500

  • 3090 10496, 24gb ram - $1500-1900

  • 3090ti 10752, 24gb ram - $1500-1900

  • 4090 16384, 24gb ram $1700-2400

Above are listed by cuda cores low to high.
4080 is the worst but has the weird bump in ram.
The 3090 is a happy medium, but you might as well get the 3090ti as the price is close/could be same.

But at that point a couple of hundred dollars more you could find a 4090 with 6000 more cores and the same ram.

As long as your power supply and case is large enough, I’d say the 4090 is the wiser more future proof purchase. And not much more money.

Edit: typo. fixed order.

153

Now that I’m reading this I think I’ll return the 4080 and buy a 4090…

1 Like
154

Just remember they are hungry cards. Lots of power draw.

155

I know. I have now 850w but I think I will need at least 1000w or maybe even 1200w.

156

thank you.
Based on your advice, I would choose a gpu.

157

Hi!

I had HQPlayer Embedded working well a year (few years?) ago, but fell out of the habit of using it.

I want to blow the dust off my configuration and give it all a good listen again, so I:

  • updated the OS of the machine I’m running both RoonServer and HQPlayer on from Ubuntu 20.04 to 22.04
  • installed libgmpris_2.2.1-8_amd64 and hqplayerd_4.34.2-156_amd64

and fired up hqplayerd as user hqplayer.

The NAA I want to talk to is a Signature Rendu SE, which I’ve told to switch to its NAA personality instead of being a Roon endpoint.

I eventually found/remembered to go to [server]:8088/config to configure stuff, but I can’t seem to get a configuration using a Network Audio backend to stick.

I am seeing

NAA output discovered 0 Network Audio Adapters

in the log, which, you know, seems like it could be a clue.

So… how can I help hqplayerd not fail at NAA discovery?

Note that on the machine this is running on, there are two hardware ethernet interfaces, one talking on the 10.0.1.x network and the other talking on the 10.0.22.x network. The former is used for Roon user-interfaces to contact the server, and for the Roon server to access streaming content; the latter talks with the audio-devices-only VLAN which the Roon endpoints (and the NAA) live on.

Disabling either network is not an option, nor do I want to have to move the NAA to the 10.0.1.x general-purpose network every time I switch to trying out HQPlayer.

Is there some way to hardcode the IP address of the NAA in the hqplayerd config file (it will always have the same address; there’s no need for discovery) or, failing that, to tell hqplayerd which subnet to look to for discovery?

Or am I mis-diagnosing what’s going wrong here?

Thanks!

158

This is the issue.

NAA protocol always uses discovery, like UPnP and many other things. It doesn’t work on concept of IP address or similar. Instead it is based on concept of NAA’s name and name of the device behind it.

As usual, multicast is used for the discovery. And typically it is routed towards default route.

Why not?

No, since NAA is conceptually is above layers like IP.

159

It’s been 3 weeks… I haven’t made any changes. :grin:

Now I have too many filter choices to choose from… haha :crazy_face:

3 Likes
160

Thanks for the prompt reply earlier!

I really didn’t want to change how things were set up, because I’d achieved exactly the network setup I wanted for Roon and all its zones: the Roon endpoints living in a VLAN dedicated to just Roon audio traffic, so they weren’t being bombarded with the multicast traffic from anything not Roon-related, the Roon Core machine talking to the audio VLAN out one interface (which had no traffic not related to playing audio out to endpoints) and talking to Roon control clients and the outside world including streaming services out the separate main-LAN interface.

But as was predictable, I pulled the particular Rendu I’ll be sending HQPlayer-processed audio to out of the Roon-audio VLAN back into the general-purpose VLAN, and HQPlayer then found it.

I’m still a little itchy about that, but yes, HQPlayer Embedded is now working nicely talking to that Rendu in NAA mode. I’ve even settled into some settings I’m enjoying listening with without a lot of frenzied changing back and forth. I was happy to learn that once I realized that the AVX2 version of HQPlayer was appropriate for my CPU generation and installed that, performance was enough better that I can reliably use poly-sinc-gauss-long / ASDM7ECv2 to DSD512 x48. And that sounds pretty darned swell through my May KTE.

So thanks again for the reply (and for the offline license key update)!

2 Likes