HQPlayer Linux Desktop and HQplayer embedded

[Miska]

2 minutes ago, chipvn said:

However, I have not seen any graph showing how the ASDM7EC-light/ul or the 2 new AHMs.

 

You will have fun time trying to see any graph differences between ul/light/super. If you listen, you will likely notice a certain difference though. And depending on your source, you may prefer one of these.

 

[chipvn]

4 minutes ago, Miska said:

If you listen, you will likely notice a certain difference though.

That's the reason why I'm have been trying to achieve DSD1024 with ASDM7EC-light/ul even though there were AMSDM7EC 512+fs / AHM available.

I do remember you mentioned before that you don't see the need to have anything lighter than the ASDM7EC-ul. However, I think for a minor group of passive cooling fans, a lighter ul version will be really nice, though.

Thank you

[Miska]

4 minutes ago, chipvn said:

That's the reason why I'm have been trying to achieve DSD1024 with ASDM7EC-light/ul even though there were AMSDM7EC 512+fs / AHM available.

 

I would keep AMSDM / AHM separate from the others. These are two separate categories. But EC-ul/light/super are same category bug different. And I would encourage to listen to these three variants without preconceptions about what is "better". Because it is not that straightforward. One may prefer for example "light" over "super" depending on music genre and personal preferences. I personally prefer ASDM7EC-light on the Gustard A26.

 

4 minutes ago, chipvn said:

I do remember you mentioned before that you don't see the need to have anything lighter than the ASDM7EC-ul. However, I

think for a minor group of passive cooling fans, a lighter ul version will be really nice, though.

 

"ul" is already pushed to the limits. Overclocked passive cooled RPI5 can do DSD256 with ASDM7EC-ul. It could be possible to shave off some percents further. But I'm not expecting any notable differences.

 

And from the other end, development of CPUs over time will make these relatively lighter. For example things are quite different now compared to the situation in 90's when I started this project... Both algorithms and the computing hardware...

 

[chipvn]

1 hour ago, Miska said:

And from the other end, development of CPUs over time will make these relatively lighter.

Totally agreed on this.

 

1 hour ago, Miska said:

I would keep AMSDM / AHM separate from the others. These are two separate categories.

From technical standpoint, which one is the better one at high rate (like DSD1024) between AMSDM7EC 512+fs and AHM7EC5L?

Thanks

[Gato]

hqp5-control related question:

• When: hqp5-control <HQPlayer Desktop 5 ip address> --get-modes
  • The system respond:
    • [0] "[source]" -1
    • [1] "PCM" 0
• When: hqp5-control <HQPlayer Embedded 5 ip address> --get-modes
  • The system respond:
    • [0] "[source]" -1
    • [1] "PCM" 0
    • [2] "SDM (DSD)" 1

What would be the reason for this?

Thanks for helping!

[Miska]

8 hours ago, chipvn said:

From technical standpoint, which one is the better one at high rate (like DSD1024) between AMSDM7EC 512+fs and AHM7EC5L?

 

Depends how you look at it, and what kind of DAC is being used, maybe AMSDM7EC. But usually from technical standpoint DSD256 or DSD512 performs better than DSD1024.

 

[Miska]

1 hour ago, Gato said:

hqp5-control related question:

• When: hqp5-control <HQPlayer Desktop 5 ip address> --get-modes
  • The system respond:
    • [0] "[source]" -1
    • [1] "PCM" 0
• When: hqp5-control <HQPlayer Embedded 5 ip address> --get-modes
  • The system respond:
    • [0] "[source]" -1
    • [1] "PCM" 0
    • [2] "SDM (DSD)" 1

What would be the reason for this?

Thanks for helping!

 

In former case the DAC doesn't support DSD?

 

[Gato]

19 minutes ago, Miska said:

 

In former case the DAC doesn't support DSD?

 

Right, thanks!

PS:

In fact, there is no DAC connected to "HQPlayer Desktop 5", there is only a DAC connected to "HQPlayer Embedded".
However, the fact that on "HQPlayer Desktop 5" under "Preferences", I could see the SDM tab, even when no DAC is connected, "misled" me.

[StreamFidelity]

18 hours ago, chipvn said:

The 2 new AHM modulators do not sound good to my preference. They are a bit less details and less dynamic although they just need about 3.9GHz max CPU clock for DSD1024.

 

I can't confirm that. In the meantime, I tried ASDM7EC-super 512+fs and ASDM7EC-light 512+fs again. And I went back to AHM7EC5L after all. The bass and space are better. Basically, my first impression was confirmed again. 

On 11/28/2023 at 5:57 PM, StreamFidelity said:

I was very excited about the new AHM7EC5L modulator. I am running a fanless system with i9-13900K with 8 P-Cores on max. 4.4GHz and 2 E-Cores enabled. Matrix Convolution is activated. 

 

Less power is required than before with DSD512 and ASDM7EC-super 512+fs. My first impression is a cleaner, deeper bass and more space. I think I can hear more micro details. Great work!

 

Or here from another user. 

On 11/29/2023 at 2:26 PM, Kalpesh said:

IMO it beats Super @256 or Light @512 with focus on clean deep bass extension and imaging

 

But you can't generalize. It depends on the DAC and the other hi-fi components. 

[chipvn]

4 hours ago, StreamFidelity said:

But you can't generalize. It depends on the DAC and the other hi-fi components. 

Agreed. It is still system and preference or sometimes even listening mood dependence.

Having more available configurations provides more choices for all of us.

I might end up using DSD512 mainly and DSD1024 occasionally...

[ZaphodStyle]

On 4/8/2024 at 8:19 AM, Miska said:

AMSDM is quite different from ASDM. So these are two completely different modulator categories, although the naming difference may not be obvious. AMSDM is similarly very different things as are the newer experimental AHM modulators. These work generally well on discrete DSD DAC implementations, within their constraints, as long as the output rate is high enough.

Miska,

   Coulkd you please clarify what you mean by "discrete DSD DAC implemention"? I am using an iFi iDSD2 with its Burr-Brown multibit DAC implmention. Is this an example of a discrete DSD DAC please?

[bogi]

@ZaphodStyle Discrete DAC means no DAC chip but discrete components used for D/A conversion.

 

https://www.tme.eu/ie/en/news/library-articles/page/51364/discrete-components-characteristics-and-examples/

A discrete component can be defined as a single part of an electrical circuit with one dominant function. It can be for example a resistor, a capacitor or a diode. Discrete electronic components can be passive or active. They are also different in the way in which they are mounted – they are designed as THT components or small SMD components.

The division into discrete and integrated components is mostly intended to separate singular components from integrated and hybrid circuits, inside which many different functions are usually performed.

In DAC world 'discrete DAC' relates only to D/A stage implementation itself. Other parts of DAC device implementation like USB and SPDIF receiver, remote control etc. are usually implemented with integrated circuits.

 

Discrete DSD DAC implementation examples are Holo DACs, T+A DAC 200, Gustard R26, Denafrip DACs, TEAC UD-701N.

[Sunny_Player]

@Miska

Thanks for providing the 5.5.2 embedded update, everything works fine on my side. Just a quick one: Could I ask you to explain what "pre before metering" actually does? I do not get it out of your explanation. Thank you. 

[sledwards]

6 hours ago, Sunny_Player said:

@Miska

Thanks for providing the 5.5.2 embedded update, everything works fine on my side. 

I loaded 5.5.2 embedded and the Matrix Pipeline still shows 128 lines when I have selected to use 8 DSP pipelines on the configuration page. Restarted/rebooted my HQPlayer machine with the same results. Not a show stopper, but I thought I would report my findings to you.

[Miska]

6 hours ago, Sunny_Player said:

Thanks for providing the 5.5.2 embedded update, everything works fine on my side. Just a quick one: Could I ask you to explain what "pre before metering" actually does? I do not get it out of your explanation. Thank you. 

 

I have tried to explain it in the manual / help-page. Metering is the spectrogram + level meters feature in for example Client, where you can monitor the input signal. Then there are pre-processing functions like the 20 kHz filter which you can easily toggle on the fly from Client (now in 5.6.0 much easier). By default, metering always shows the source content, exactly as it arrives in HQPlayer. But this option allows you to run the pre-process first, before feeding it to the metering process. This allows you to see the effect of 20 kHz filter in both spectrogram and the level meters.

 

However, usually 20 kHz filter is toggled based on the spectrogram display, and if "pre before meter" is enabled, you are kind of "blind" in a way that you don't see when you could turn it off...

 

So both ways are useful in their own ways, thus the configuration option.

 

[Miska]

39 minutes ago, sledwards said:

I loaded 5.5.2 embedded and the Matrix Pipeline still shows 128 lines when I have selected to use 8 DSP pipelines on the configuration page. Restarted/rebooted my HQPlayer machine with the same results. Not a show stopper, but I thought I would report my findings to you.

 

Ahh, I forgot to fix this because I was rushing to make this release. I found one case where HQPlayer could crash in certain situation and wanted to get fix out for that. Also Nvidia quickly fixed some bugs in CUDA between the 5.5.1 and 5.5.2 and I wanted to have those included as well.

 

This is visual bug that doesn't effect processing behaviour (extra pipelines are just ignored), but of course it is supposed to only show the active ones and is more practical that way. So it will be fixed.

 

[StreamFidelity]

@Miska 

HQPlayer 5 Embedded 5.5.2 does not work for me. I have applied the previous settings with Restore, so there should be no setup error.

 

The core distribution is completely confused. RAM shoots up.

I have gone back to HQPlayer 5 Desktop 5.5.0 

Everything is fine there.

[chipvn]

5 minutes ago, StreamFidelity said:

@Miska 

HQPlayer 5 Embedded 5.5.2 does not work for me.

I think he mentioned that he found the issue that make Sinc-MG(a) works but Sinc-Mx dropouts in many system.

He correct the issue in this new release but it means from now Sinc-MG(a) will also dropout in system that Sinc-Mx dropouts.

😅

[StreamFidelity]

1 minute ago, chipvn said:

I think he mentioned that he found the issue that make Sinc-MG(a) works but Sinc-Mx dropouts in many system.

Many thanks for the information. Unfortunately, there was also a malfunction with poly-sync-gauss-xla. Roon was able to connect to HQP, but the progress bar did not move and no music played. The cores were not loaded. 

[chipvn]

HQPE OS 5.5.2 works fine for me and performance is the same as 5.5.1.

There is 1 minor issue I found is filter / modulator could still be changed real-time via web interface. However, those new filter / modulator will not be updated in Roon status. Roon still shows the filter / modulator that was set in configuration.

[chipvn]

7 minutes ago, StreamFidelity said:

Unfortunately, there was also a malfunction with poly-sync-gauss-xla. Roon was able to connect to HQP, but the progress bar did not move and no music played. The cores were not loaded. 

Poly-sinc-gauss-xla works fine for me at 1x from 24/48 to DSD512x48 and ASDM7EC-super 512+fs. Multicore enable, ecores filter, nblocs default.

 

Edited: might be you could try resetting from fresh instead of restoring your backup setting and see if it works

[Sunny_Player]

28 minutes ago, chipvn said:

Poly-sinc-gauss-xla works fine for me at 1x from 24/48 to DSD512x48 and ASDM7EC-super 512+fs. Multicore enable, ecores filter, nblocs default.

 

Edited: might be you could try resetting from fresh instead of restoring your backup setting and see if it works

MG (a) Filter is not working for me.

[erekola]

4 minutes ago, Sunny_Player said:

MG (a) Filter is not working for me.

@jussi_laako said he found a bug in the sinc-MGa, now it really requires a lot more power!

 

 

[Sunny_Player]

1 hour ago, erekola said:

@jussi_laako said he found a bug in the sinc-MGa, now it really requires a lot more power!

 

Thank you. What gpu do you use?

[erekola]

1 hour ago, Sunny_Player said:

 

Thank you. What gpu do you use?

CPU: i9-14900KS GPU: RTX4090 RAM: 48gb 8400MT/s