HQ Player

[botrytis]

Miska, I just wanted to ask your thoughts on Apple going to their own CPU/GPU's. They are ARM instruction set but their own design for the CPU. The GPU I am not sure of. Can an ARM CPU run this?

[LoryWiv]

20 hours ago, Miska said:

 

... try setting DAC Bits to 20 or so. It is better to play safe and set the number of bits a bit too low rather than the other way around. 

 

Thank you @Miska. More generally, I had impression that for most use cases one should set DAC bits as high as the DAC supports. Does that still hold true and if not, what determines optimal setting for a particular DAC?

[luisma]

19 hours ago, StreamFidelity said:

I love the sound of the PCM / Sync-L

I know you have a Terminator like me and you mentioned 2 days ago DSD 256 7EC was your preference and I did not disagree but personally I love PCM in that environment. I was just wondering if I could be so wrong on preference but it seems the R2R DACs thrive with PCM.

[StreamFidelity]

29 minutes ago, luisma said:

I was just wondering if I could be so wrong on preference but it seems the R2R DACs thrive with PCM.

 

No you are not wrong. The T+ has made a significant leap forward in this regard. Today I made a comparison with ASDM7EC / DSD 256 / sinc-M (sinc-L doesn't run without GPU). Yes, ok, there was a pinch of more details to be heard. But PCM may have the nicer timbres. The T+ must be well burned in. There is still something going on. 👍

[Miska]

26 minutes ago, LoryWiv said:

Thank you @Miska. More generally, I had impression that for most use cases one should set DAC bits as high as the DAC supports. Does that still hold true and if not, what determines optimal setting for a particular DAC?

 

"Default" gives you the highest "smart" figure. In certain cases it should be limited further.

 

For example cases where S/PDIF or AES/EBU interface is involved, it should be set to max 24-bit, or lower if the DAC has less resolution. Because those interfaces are unidirectional, HQPlayer cannot figure out if a DAC connected behind such interface is for example 16-, 18- or 20-bit instead of 24. Otherwise the extra bits will get lost in truncation which in turn causes distortion.

 

Holo Audio Spring (2) correctly reports 24-bit resolution despite 32-bit samples, but this is not reflected through most drivers like ASIO or CoreAudio on macOS. On Linux it works correctly with new enough kernels. Thus in other cases it should be set if sending PCM there. However, the R2R ladder doesn't reach that much linearity, so for best performance DAC Bits should be set to at most 20-bit. (DAC also needs to be set to NOS mode of course)

 

For other R2R DACs, they pretty certainly don't have 24-bit worth of linearity, so it is better set the DAC Bits to at most 20, or less.

 

If the DAC has built-in DSP, like modern DAC chips, then it is better to give the DSP as much precision as possible, most of the time meaning 32-bits, which is what you get with "Default" setting in such cases. Example of such would be for example RME ADI-2.

 

This setting naturally only applies to PCM output mode.

 

[Miska]

2 hours ago, botrytis said:

Miska, I just wanted to ask your thoughts on Apple going to their own CPU/GPU's. They are ARM instruction set but their own design for the CPU. The GPU I am not sure of. Can an ARM CPU run this?

 

Remains to be seen what kind of performance those have. Based on other ARM chips I've used I'm not holding my breath.

 

They will certainly have fun time designing something like iMac Pro or Mac Pro based on ARM. But we will see...

 

[luisma]

13 minutes ago, StreamFidelity said:

But PCM may have the nicer timbres

And that is EXACTLY what I experienced too, so glad to find someone with kind of the same "SQ evaluation profile", the timbre (I would say tonality myself but kind of the same thing just the way it is applied) is spot on, it recreates the illusion which is critical for the experience.

Other DACs (DS) very possibly perform better with DSD, my project S2 does

[luisma]

13 minutes ago, Miska said:

DAC Bits should be set to at most 20-bit. (DAC also needs to be set to NOS mode of course)

Learned from you this long time ago, I use 19-bit

[ted_b]

38 minutes ago, luisma said:

And that is EXACTLY what I experienced too, so glad to find someone with kind of the same "SQ evaluation profile", the timbre (I would say tonality myself but kind of the same thing just the way it is applied) is spot on, it recreates the illusion which is critical for the experience.

Other DACs (DS) very possibly perform better with DSD, my project S2 does

Could I implore you guys to please tell us the source file format when comparing upsampled PCM to remodulated DSD. It is much more a news item if DSD files have been down converted to Very high PCM and yet still have better timbre than when remodulated at, say, DSD256   In other words, I don’t find it unusual at all when R2R dacs present pcm files upsampled to very high PCM better than when PCM files are converted to anything DSD. My $.02

[Miska]

49 minutes ago, luisma said:

And that is EXACTLY what I experienced too, so glad to find someone with kind of the same "SQ evaluation profile", the timbre (I would say tonality myself but kind of the same thing just the way it is applied) is spot on, it recreates the illusion which is critical for the experience.

Other DACs (DS) very possibly perform better with DSD, my project S2 does

 

Many DACs like Holo, Denafrips and T+A have two separate converters, one for PCM and another one for DSD. Then rest is about which of the two DACs perform better when driven in optimal way.

 

[StreamFidelity]

1 hour ago, ted_b said:

Could I implore you guys to please tell us the source file format when comparing upsampled PCM to remodulated DSD.

 

In my case, the source is always PCM. My observations relate to 44.1 / 88.2 and 48/96.

 

As Miska has already written, separate converters are implemented in the Denafrips Terminator-Plus. Both are excellent in my opinion:

 

- True balanced 26BIT R2R (PCM)

- 6BIT DSD (32 steps FIR filters)

 

For me it is a matter of taste. I am happy that I can change whenever I want. 👍

[luisma]

2 hours ago, ted_b said:

In other words, I don’t find it unusual at all when R2R dacs present pcm files upsampled to very high PCM better than when PCM files are converted to anything DSD. My $.02

You're correct Ted, source is PCM not DSD with the testing I have done

[luisma]

5 minutes ago, StreamFidelity said:

As Miska has already written, separate converters are implemented in the Denafrips Terminator-Plus. Both are excellent in my opinion:

Yeah but IMO the PCM stills wins, feeding DSD to the Terminator sounds terrific but I might be biased towards PCM, I tested this a lot trying to prefer DSD (since Miska vouches for it most of the time and I respect a lot the work he has done with digital from the moment I heard HQP for the 1st time)

I have tube preamp and amps too, this could influence as well

 

[Gavin1977]

23 hours ago, StreamFidelity said:

Jussi, thank you very much for the information.

 

Just for control. When upsampling a CD resolution (44.1 / 16) to 1,411.2, are we talking about 4,194,240 (131070 x 32) tabs? Sensational! That is four times as much as a CHORD HUGO M SCALER.

 

I love the sound of the PCM / Sync-L. I'm not sure yet whether it will top DSD with ASDM7EC. But it's getting dangerous. 😄

 

For the CPU, the load of 1-7% is a piece of cake.

 

Great feedback - I have T+A DAC8 DSD, so making me consider switching to Denafrips to do up sampling in PCM due to lower CPU requirements.  Good to know it’s close.  Further comparison would be great!

 

I don’t know what CPU clock rates would be required for Sinc-L, DSD512 and EC modulator.  I would consider water cooling if it was achievable as I’m keen to avoid GPU.

[luisma]

1 minute ago, Gavin1977 said:

I don’t know what CPU clock rates would be required for Sinc-L, DSD512 and EC modulator.  I would consider water cooling if it was achievable as I’m keen to avoid GPU.

I don't think this is possible with current existing CPU's

[Miska]

Only with offline conversion with HQPlayer 4 Pro. That way also DSD1024. But it is a lot of effort and needed storage space to do it...

 

[botrytis]

20 hours ago, Miska said:

 

Remains to be seen what kind of performance those have. Based on other ARM chips I've used I'm not holding my breath.

 

They will certainly have fun time designing something like iMac Pro or Mac Pro based on ARM. But we will see...

 

 

Thanks for your reply - it is always appreciated from someone on the front lines  

[Yviena]

@Miska Would you recommend to leave dac bits at 32 for AKM/ESS dac's in PCM mode,  or use 21-22 bits for linearity with volume control done via analog pot?

[il Carletto]

Hello,

I'm using HQP Desktop on my new pc (Intel 10th Gen. i7-10700K OC @4800). No GPU.

OS is Gentooplayer Xfce 2.15. My Naa is an Allo Usbridge Sig (Gentooplayer as well).

Dac is a RME Adi-2 fs

 

I am experiencing many problems at DSD256 with no really heavy setting (poly-sinc-lp-2s + ADSM7) and an abnormal cpu resources use (more than 80%). It doesn't matter the file I'm playing.

Gentooplayer is running on a fast USB 3. No other programs installed. 

 

My computer should be powerful enough to play with this setting, at least.

Any hint?

Thanks in advance

[luisma]

Just be sure you are not using core isolation, turbo and HT must be on. I had similar issue months ago and from what I remember it was the use of profiles, core isolation etc. Leave everything as it comes on default install and test, then if you want start adding optimizations by all means but IME with GP Hqplayer will benefit from maximum power and resources.

 

[luisma]

I see only one Cpu (physical) being used? Most likely profiles, sometimes you have to erase it manually, let me get my notes

[il Carletto]

Thank you Luisma, I will check asap.

I'm only using XMP1 for overclocking, but I don't really know that bios very well yet.

[il Carletto]

I tried to run HQP on Ubuntu 20.04, but no success.

I still have same problems.

[ckpiv]

9 hours ago, il Carletto said:

Hello,

I'm using HQP Desktop on my new pc (Intel 10th Gen. i7-10700K OC @4800). No GPU.

OS is Gentooplayer Xfce 2.15. My Naa is an Allo Usbridge Sig (Gentooplayer as well).

Dac is a RME Adi-2 fs

 

I am experiencing many problems at DSD256 with no really heavy setting (poly-sinc-lp-2s + ADSM7) and an abnormal cpu resources use (more than 80%). It doesn't matter the file I'm playing.

Gentooplayer is running on a fast USB 3. No other programs installed. 

 

My computer should be powerful enough to play with this setting, at least.

Any hint?

Thanks in advance

Is your music not playing correctly?

 

Your load average is very minimal (.37 1min .30 5 min .14 15 min)l. If not mistaken the i7-10700k you have is 8 cores and 2 threads per core,  so your CPU % Max would be 1600%, you need to factor in how many cores and hyper threads per core. Looking at your screen shot it appears your system is hardly working. Others with more knowledge can chime in. 

[austinpop]

I'm trying to understand clipping. Are there any scenarios where the volume knob turns red, but the "Limited" count stays at 0? Shouldn't the red knob correspond to a non-zero count?

 

This is with PCM to PCM, sinc-M, LNS15, 768k.

 

HQPlayer 4 Desktop 4.60, Windows 10.