HQ Player

[volpone]

2 hours ago, bogi said:

I bought this one for 77 EUR with Win10 preinstalled. For test purposes I installed also Ubuntu and can run NAA on both OS.

 

Hi Bogi, did you tried "this one" to stream DSD512 on Win10 with a good SQ ?
Any enhancement replacing PSU ?
 

I'm using T+A DAC8DSD and due to Amanero Linux driver limitation i'm unfortunately sticked on Windows ASIO.  I'm looking for a good win NAA endpoint HW.

Thank you.

[volpone]

16 minutes ago, denniswaugh said:

I just hope a solution is found to overcome the Amenero/Linux issue that would enable DSD512 playback through the MicroRendu.

 

So do i ... but unfortunately no move on this issue since quite one year. Sad !   

[volpone]

2 minutes ago, hifi25nl said:

It is just a line of code in the kernel... a recent version of alsa and the firmware 1099be on the Amanero

https://github.com/lintweaker/xmos-native-dsd/blob/master/SRPMS/patches/kernel/0001-usb-Add-native-DSD-support-for-Amanero-Combo384-BE.patch

 

 

 

Thank you @hifi25nl , i've asked T+A and they told me it was on the Amanero side and they were "pushing" ...

But nothing new .... who is waiting who or what ? 

[volpone]

1 hour ago, hifi25nl said:

Linux is ready. The problem is that the Amanero driver is not working fine, for long time in beta stage...

 

Any idea on how to help ? 
 

I've read the discussion from: https://github.com/lintweaker/xmos-native-dsd/issues/12 but don't know how to "motivate" @amanero and / or the community ?

 

If this issue fixed i think more DACs and cards would be sold ...

 

[volpone]

44 minutes ago, denniswaugh said:

My Tronsmart Ara 5 remains cool to the touch.

The only issue I had with it was the built in ethernet had limited bandwidth during DSD512 playback and would clip.

This issue was resolved by using a Realtek  usb3 to ethernet adapter. No problem with DSD512 playback.

 

I am only using this Win10 NAA solution because it is the only way I could listen to DSD512 with my Lampizator dac.

 

I still prefer the sound with the MicroRendu/LPS-1 but this is limited to DSD128 at the moment. 

 

Cheap and small Win NAA HW !
 

Are you comparing  DSD512 (ARA 5) VS DSD128 (MR + LPS1) and prefer SQ from MR in this context ?

Did you tried to update ARA PSU ?

[volpone]

29 minutes ago, denniswaugh said:

....

I suspect there is a limit to what can be achieved with the Ara5 and similar hardware platforms because you cannot readily modify or optimise the hardware.

Annoying if you are right Dennis ...

Lot of ressources required on the DSD128  > DSD512 path and if no real SQ benefits at DAC outputs, wasted ressources !
 

Hope for some better Win NAA HW platform or Linux driver fix ! But perhaps i'm dreaming and the best way is  to stay DSD128 on Linux based optimized hardware platform ? I'm currently using RPI3 NAA (DietPi) to feed my DAC8 DSD with DSD128 and SQ is already quite good.

[volpone]

15 minutes ago, Miska said:

 

That is rather poor version, mine supports both BE and LE firmware variants in the same patch and doesn't make stupid assumptions that all firmware versions support the same native DSD functionality...

 

So @Miska considering your patch what is missing from Amanero, the community, or DAC makers, to release long waited native DSD Linux driver (DSD512 enabled) ? 

[volpone]

8 minutes ago, Miska said:

 

......

What is missing is a firmware version that has working DSD indication, MUTE and DSD512...

 

 

OK, so who is in position to do the job on these 3 points, Amanero ?

A lot of work ?

[volpone]

7 minutes ago, Miska said:

 

Amanero yes, nobody else can do it. I don't know how much of work it is since I don't have any visibility. But the demands for the BE support weren't probably helping the cause... So time was spent on doing something that mpd should have done in their application instead of trying to come up with a functional firmware.

 

Did you, and DAC vendors, talked to Amanero ? Everybody agree on what to do ?

[volpone]

4 hours ago, bibo01 said:

When I talked to Amanero a couple of weeks ago, he said that his customers had a DSD512 working version of his firmware. He just had to implement a certain feature (I don't remember exactly what). He seemed VERY close to a final release. Unfortunately, he was also just about to move to a new office. I'll call him on Monday to get an update...if he still has the same number.

 

So perhaps some hope. When i talked to DAC vendor (T+A, 15 days ago) they said: 

 

" Currently Amanero is working on a solution for Linux. 
There are some discussions about HQ player + Linux on the internet which you might like to follow.
As soon as the Linux support for Amanero is ready and tested, we will implement it.
However, a fix date for this can not be given at this moment."

 

If Amenero succeed the next question would be: enduser DAC FW update possible and allowed from vendor  ?

 

 

[volpone]

2 minutes ago, bibo01 said:

As promised, today I talked to Amanero developer about his ASIO driver and firmware for Linux.

 

Thank you @bibo01.

I've got a mail exchange today with T+A and it seems they are "in the starting blocks". I really hope some move soon.  

[volpone]

Hi all,

Using HQP/NAA since few days (evaluation version). Very satisfied about SQ in my setup. However, sometimes, i hear very small and sporadic "dropouts" ...

I eventually could live with these "dropouts" because they are uncommon, sporadic, and not very disturbing, but better to fix if possible.

Any idea what is the cause ?

Thank you.

WHAT
- Small "dropouts" (sorry if this is not adequate wording), very short and "dry"
- Look like dust or physical defect on a vinyl track , 
- Nothing loud nor harsh, not very disturbing but sometimes hearable 

WHEN
- Not "clocked", very erratic few minutes without any dropouts ...

I've never heard these "dropouts":

- When i'm playing directly native DSD128 (DFF)
- When i'm playing the same streams directly from Roon to DAC without HQP in the signal path

Setup:
- HQPlayer Desktop 3.16.0 (+Roon server build 209): NUC6i5SYH WIN10
- NAA 3.5: RPI3 DietPi + IFI iPower PSU
- DAC: T+A DAC8 DSD

- Gigabit wired network

HQP Settings:
(TIDAL HIFI) FLAC 16/44.1 > DSD128
poly-sinc mp 2s, ADSM5 > DSD (DoP)

 

CPU load

- HQP NUC CPU Load (i5): < 60% (WIN10)

- NAA load < 1%

 

[volpone]

13 hours ago, simonklp said:

I understand what you meant by the sound when running HQP: "Look like dust or physical defect on a vinyl track , 
- Nothing loud nor harsh, not very disturbing but sometimes hearable"

 

I refer it as some slight "pop" or "click" noise. It can be eliminated by changing the buffer time setting from "default" to other milliseconds setting. I had asked Miska before, and he personally use 100ms and implied that it is a good starting point. His explanation of the "pop" noise is as follows: -

"the pop happens because computer didn't deliver more audio to the DAC in time and the DAC run out of data to play. Longer buffer helps keeping up with delivery deadlines."

 

I had also asked him whether smaller buffer time is better or not. But he said that it is just the other way round. His reply was as follows: -

"longer/larger buffer reduces computer activity and runs things at more relaxed pace. But apart from avoiding those drop-outs it shouldn't have any impact on sound quality."

 

However, I sometimes find setting the buffer to 100ms still have the "pop" noise, and setting it to smaller values seems to sound slightly better to me, although it is actually not much difference and may  be just my psychological feeling only.

 

Simon,
Thank you very much for you detailed and useful explanations.

I will try with buffer time adjustments. Actually my HQP buffer time is set to "default" and i don't know what is the corresponding value. Not easy to manage because the "clicks" are very sporadic.
 

Regards 

[volpone]

15 hours ago, simonklp said:

I think the advice from Miska is the most authentic and authoritative. Let's see what he says.

 

 

@Miska advice: 

"Another source of clicks is possible overloading/clipping. So make sure HQPlayer volume control is set to at most -3 dB and that the "Limited" counter in main window stays zero. In 3.16+ versions the volume knob will also turn red whenever there is overload in the pipeline, meaning the limited counter increments and the internal limiter is triggered. So it is easier to notice..."

 

I've set Vol output: -3db in HQP settings and it seems my vintage vinyl "clicks" disappeared. To be confirmed with extensive tests from different files. 

 

[volpone]

2 hours ago, luchoh said:

....

With the above settings, 3.16.0 doesn't have the vinyl noise while 3.16.1 always has the vinyl noise. I tried all possible buffer settings: from Default to 250. ....

 

Hi @luchoh,

From my side (DSD128 , HQP 3.16.0) i've fixed the "vinyl noise" with HQPlayer settings Vol Max: -3db

I've not tried yet with 3.16.1.
 

I'm using T+A DAC8 DSD and according to @Miska "Amanero drivers use fixed buffer size that cannot be adjusted anywhere. So no matter what is set in HQPlayer settings the buffer is always same size".

[volpone]

Hi all,

I’m using HQPlayer (with ROON UI) to convert and upsample sources from 16b/44.1 Khz to DSD 128 or DSD 512.

I want to apply convolution filters for Digital Room Correction purpose.

What is best compromise regarding filters files format (bitdepth, sample rate) , my concerns are about sound quality and computer ressources optimization ?
 

Thank you.

[volpone]

 

3 hours ago, Miska said:

Convolution filter file bitdepth and sample rate generally don't have notable effect on computer resource usage, apart from the initialization time before playback begins.

 

If you use Acourate, then I'd recommend to generate filters for 384k sampling rate, 64-bit mono WAV. Remember to check the "Brick" box too, then you can keep the "Expand HF" off in HQPlayer - then it needs to do less massaging on the filters...

 

Thank you @Miska.

If my sources files are only 16/44 and output only DSD 512 do you still recommend "to generate filters for 384k sampling rate, 64-bit mono wav" ?

[volpone]

Hi all,

Any compute performance comparison using HQPlayer (upsampling + convolution) with Intel new Coffee Lake I-8700  (6cores, 12 threads) VS AMD Ryzen R1700 (8 Cores, 16 Threads) , both TDP 65W ?

[volpone]

2 hours ago, Miska said:

Support for Amanero is in latest Linux kernels, but it also requires fresh firmware from Amanero. Stock kernels tend to be too old, so I'm building custom ones...

 

Hi, my question is a little OT but not so much ...

Anybody tried to flash "fresh" AMANERO firmware on T+A DAC8 DSD ?

- Witch FW version ?

- Any audio glitches (DSD512 native) ?

Thank's

[volpone]

On 30/10/2017 at 5:36 AM, Argopo said:

1. Small pop (more of a single crackle) before the start of a track if playing back a DSD file.

2. Faint overlay of static above the music when playing back hi-res (24/96, 24/192) PCM files upsampled to 512. The static overlay is not present when upsampling 16/44.1 files.

 

The static overlay is present when using Roon as an interface or when using HQPlayer standalone.

 

Control PC is standard 6700k 16GB build.

Windows-based NAA.

T+A DAC 8.

 

Hi, @Argopo

As you know T+A DAC8 DSD use Amanero USB.

Seems new Amanero ASIO driver is on the way to be certified (fingers crossed).
Perhaps some enhancements for your setup. I'm also waiting for the Linux version without glitches. See: 

 

[volpone]

On 2/5/2019 at 6:51 PM, ddetaey said:

But, I do prefer non-2s as well. 

With my T+A Dac I even slightly prefer DSD256 - poly-sinc-xtr-lp compared with DSD512 - poly-sinc-xtr-lp-2s

I wish I understood what is different between 1 step and 2 steps in a particular filter, as the Dac itself is not aware what happens with the bits before receiving them.

...

 

Hi Dirk,

Interested on your feedback with T+A DAC. 

- Are you using AMD RYZEN CPU and which model ? 

- What is your CPU load for DSD256 - poly-sinc-xtr-lp  ? 

- Which associated modulator (ASDM7 or other ?) and do you use convolution filter too ?

Thank you ...

[volpone]

Thank you @ddetaey,

Dirk, one more question: Based on your experiences what about SQ differences between DSD256 and DSD512 using DAC8 DSD ? Filter dependent ?

[volpone]

3 minutes ago, craighartley said:

And I assume this is partly because it is only when you feed it 512 that you are taking full advantage of the T+A DAC8 DSD’s filterless DAC. As I under stand it, when you feed it 256 the DAC has to do extra processing. So anyone preferring 256 to 512 using the same filter in HQPlayer is actually preferring the effect of the DAC’s extra processing. 

 

Interested in technical insight from @Miska on DAC8 DSD extra processing when fed from DSD256 ...
Also I think he made some measurements comparing DSD256 vs DSD512 ?

[volpone]

5 hours ago, ddetaey said:

 

To be honest, for me filter is more important than upsampling rate, at least with the T+A 8 DSD dac.

To be complete, I do prefer the heavier filters, and as said before, I prefer DSD256 with a single step filter (non-2s) than DSD512 with 2-step filter (-2s) Due to this preference I have not tested the lightest filters such as poly-sinc-short, neither in DSD256 nor DSD512.

 

Please take into account that DSD512 on native Linux does not work with the T+A 8 DSD Dac (or any other dac using the Amanero USB module). Apparently it works fine using Windows but I have not tried that (nor am I interested to try that).

On an ifi Micro DSD dac, DSD512 is working fine (with a poly-sinc-xtr-lp-2s filter in my case, as my streamer processor cannot handle the load of the 1-step filter). However, for me there is a hugh SQ difference with the T+A 8 DSD, which is much better even with DSD256.

 

To conclude, for me DSD256 is as good as DSD512 for what I can hear, and due to my fanless streamer, this matches nicely with my streamers capapilites of handling Jussi's top filters....

 

Thank's Dirk for your detailed answers. I've too updated AMANERO USB receiver FW to 1099c and i'm pleased to use my DAC from LINUX device (NAA under DietPi on ALLO USBridge) up to DSD256 (native):  Good SQ and very stable. I'm probably going to take RYZEN 2700 way to get enough CPU "horse power" to run HQP "non -2S" heavy filters. 

Best.

[volpone]

20 minutes ago, PieterP said:

Why not wait a few months for the new ryzen 3000 series? Even more ooomp per euro and per watt tdp. 

 

Sure, will see ! i'm looking for good compromise ooomp / TDP / euros. 2700 CPU power (65W TDP) seems already quite OK.