HQ Player

[craighartley]

7 hours ago, Miska said:

AFAIK, the DAC8 DSD is bit-perfect at all DSD rates. I would say it is matter of taste whether one prefers it at DSD256 vs DSD512... In that respect it is a bit like Holo Spring too.

 

There is similar difference also with choice of analog filter, "Clean" vs "Wide".

This exchange is what I was remembering imperfectly:

 

 

Posted April 11, 2016

  Superdad said: 

Hi Ed:

 

John hypothesized to me the other night that part of the reason the T+A DAC sounds so much better at 512 is that is run at the frequency of the clock. The act of dividing down the master clock itself produces additional jitter.

 

Oh that's right! That would actually make sense as to why its so disproportionally better. Dsd64 to dsd128 to dsd256, based on noise profiles, sure. But the big step up to dsd512? I think you may have hit the nail on the head here!

[Miska]

14 minutes ago, craighartley said:

 

John hypothesized to me the other night that part of the reason the T+A DAC sounds so much better at 512 is that is run at the frequency of the clock. The act of dividing down the master clock itself produces additional jitter.

 

Oh that's right! That would actually make sense as to why its so disproportionally better. Dsd64 to dsd128 to dsd256, based on noise profiles, sure. But the big step up to dsd512? I think you may have hit the nail on the head here!

 

Yeah, no processing difference, but a difference in clocking. I don't seem to have saved all Jtest plots, so cannot quickly check. But as I said, this goes in to matter of taste category, which parameter you want to emphasize...

 

[bogi]

On 2/3/2019 at 1:40 PM, luisma said:

Old and off topic for this thread but @bogi I have the exact same issue, did you ever found out why the NAA could not get out of the loop?

Hi @luisma , to the topic of Linux NAA discovery on Windows HQPlayer computer in the case of direct ethernet cable connection (without network switch):

I desribed my solution is here

And my sound comparison of Linux based and Windows based NAA here:

 

Because of Windows 10 NAA sounds better to my ears I am using that one.

When I wrote those posts in 2017 I just moved to a new house where I didn't have any network infrastructure available so I used a USB 4G dongle and no local network at all. Later when I had both ethernet and WiFi networks available I compared the new possibilities with direct cable connection. I remembered Miska's posts where he suggested to use network switch, which adds another level of isolation between HQPlayer computer and NAA computer. Few other forum members posted information that combination of WiFi and ethernet may sound the best. My result is: I reached the best sound with [HQPlayer notebook] > WiFi > [router] > ethernet > [switch] > ethernet > [NAA]. That sounded clearly better than [HQPlayer notebook] > ethernet > [switch] > ethernet > [NAA]. Ethernet connection with switch sounded me slightly better than direct ethernet connection without switch. I am not using any LPS or iPower with my router and switch, so maybe I could yet improve my result.

[luisma]

5 minutes ago, bogi said:

Hi @luisma , to the topic of Linux NAA discovery on Windows HQPlayer computer in the case of direct ethernet cable connection (without network switch):

I desribed my solution is here

Thank you so much for detailing the solution in such a way with the references.

Miska mentioned the dual stack (4 and 6) needs for the connection so I enabled the DHCP on Ubuntu and Miska's image comes standard with DHCP so that fixed the issue with the direct connection. Miska's also mentioned the switch in the middle will improve isolation and you are confirming it although from other posts I recall seeing the switch (if not properly powered) as another noise device which could introduce EMI. So far I am directly connected, Miska's image sounds better on my NAA than my former NAA and also my Allo USB Bridge.

 

Now my design is (damn I don't have Visio installed just reinstalled my laptop) basically this

 

RuckusWIFI AP --> WIFI ---> Fanless Ubuntu i7 HQPE

|                                                                     |

|                                                                     | This is the ethernet from HQPE to NAA 

|                                                                     |

--------------------------->  WIFI ---> Fanless Ubuntu Celeron ---- USB --> Denafrips DAC

 

Basically both fanless computers connect via wireless to the router for management only and the music flows through the directly connected ethernet cable

I actually removed the WIFI from the NAA thinking that would improve SQ, still to test.

 

I am building a core i3 fanless with Windows Server 2019 Core let's see how it goes

 

thanks again, appreciate it

 

[rando]

On 2/6/2019 at 9:39 AM, tboooe said:

 

Good idea guys but I want to keep my existing LGA1151 mobo because it can do 99% of my library at DSD512 with 2 channel convolution.  It only struggles on my DSD128 files which I only have a few.  Also, my mobo (Gigabyte Z270MX Gaming 5) does not support the gen 8 CPUs.  I am only upgrading the CPU because I will be going into a new case and adding a fan so changing it will be easy and pretty cost effective.

 

I think you should run the numbers on this before dismissing it out of hand.   i5-8400 and a Z370 board could probably be had for similar money.  Unless you are overclocking like a gamer the 8400/9400 is right there and sometimes betters the 7700K at a lower TDP (ie less heat and easier fanless cooling). 

 

You might even find a Z390 board with features that appeal more.  10 Gbps USB 3.1 gen 2 and USB C ports, more PCIe lanes, ??? you missed out on with the current mb.  The key item to keep in mind is the newer processors firmware doesn't let you access the full horsepower out of the gate.  They are designed to be utterly efficient during the majority of their lifetime that is not spent at 100% processor usage.  Once it figures out you mean business, that efficiency redoubles itself to do real work using programs such as HQ Player.    

 

[juanitox]

16 hours ago, Miska said:

AFAIK, the DAC8 DSD is bit-perfect at all DSD rates. I would say it is matter of taste whether one prefers it at DSD256 vs DSD512... In that respect it is a bit like Holo Spring too.

 

There is similar difference also with choice of analog filter, "Clean" vs "Wide".

 

i was ready to update my PC with a fancy GPU to be able to play DSD512  non 2s Polysync Xtr , then Polysync Ext2 arrive and change the game.

thanks Miska for saving my wallet 😉

 

[PieterP]

Hi Miska, thanks for the HT setting advice. What other specific bios settings can you recommend; e.g. w.r.t. turbospeed (enabled/disabled), c-state, ....? I found some recommendations before in a doc by audiophileoptimizer that i used this far.  

[Miska]

1 hour ago, PieterP said:

Hi Miska, thanks for the HT setting advice. What other specific bios settings can you recommend; e.g. w.r.t. turbospeed (enabled/disabled), c-state, ....? I found some recommendations before in a doc by audiophileoptimizer that i used this far.  

 

IMO, Turbo is also useful, especially more cores you have. It helps when core load is uneven, by allowing loaded cores to run at higher clock speed. You have a total power budget of N, and when not all cores (and maybe GPU) are fully loaded, the budget can be used to clock few cores higher. On some CPU models you can see fairly high ratio between base clock and max turbo clock. Max turbo clock case is usually single loaded core case.

 

SpeedStep works the other way, allowing clocks to fall below base clock and also voltages to drop when computer is idling. That and C-states are useful if you keep computer running all the time, not just when listening music. It is not very useful to keep CPU banging at full steam the OS' do-nothing-idle-loop. Although nowadays CPUs are anyway quite clever to power down unused parts on their own and HLT instruction enters low power mode anyway. C-states and SpeedStep are usually managed by the OS' power management software module and large part of the real action is up to PMU/PMIC firmware. The software run in PMICs is quite advanced these days, for example just powering up the computer is quite complex operation orchestrated by the PMIC. You usually don't accidentally enter deeper C-states, since these go to states called "sleep" and "suspend". I think quite many laptop owners have used suspend, but maybe not as many desktop users. Disabling deep C-states don't have effect if you are not attempting to enter those modes from the OS anyway (Control Panel / Power / Choose when the computer sleeps).

 

On Windows, the selected power/performance profile define how these features are used. On Win 10 Pro you have High Performance profile and on WIn10Pro for Workstations you have also Ultimate Performance profile.

 

I just leave those features enabled and control the settings from OS side.

 

[volpone]

9 hours ago, Miska said:

 

Yeah, no processing difference, but a difference in clocking. I don't seem to have saved all Jtest plots, so cannot quickly check. But as I said, this goes in to matter of taste category, which parameter you want to emphasize...

 

 

Thank's @Miska. About 512 vs 256 comparison with DAC8 DSD what are the  "parameter you want to emphasize" you mention : Precision vs Space ? Also interesting if you could publish some Jtest plots on this DAC ... 

[Miska]

20 minutes ago, volpone said:

Thank's @Miska. About 512 vs 256 comparison with DAC8 DSD what are the  "parameter you want to emphasize" you mention : Precision vs Space ? Also interesting if you could publish some Jtest plots on this DAC ... 

 

DSD256 gives lowest level of audio band noise floor while DSD512 gives lowest level total wide band noise. That is similar to Holo Spring. Probably the net result of these aspects depends on rest of the system, like amps and such. But you get about the same amount of difference by switching the analog filter of DAC8 DSD between "clean" and "wide". So OOB noise at DSD512 when set to "wide" is about same level as OOB noise at DSD256 when set to "clean".

 

For example I see that you use class-D amp which may behave differently in this respect compared to some class-AB amp. So I would just try and see which works best in a particular setup.

 

P.S. With these DACs, going to DSD128 or DSD64 doesn't win anything, only loose. So the question is really only DSD256 vs DSD512.

 

[volpone]

2 hours ago, Miska said:

P.S. With these DACs, going to DSD128 or DSD64 doesn't win anything, only loose. So the question is really only DSD256 vs DSD512.

 

Thank's , i'm not sure i'm understanding correctly your PS.
Using HQPlayer and DAC8 DSD i heard some SQ differences going from PCM 16/44 to DSD 128 ... Do you estimate it is technically "only loose" ? However i agree DSD256 is quite better than DSD128 ...

[Miska]

26 minutes ago, volpone said:

Thank's , i'm not sure i'm understanding correctly your PS.
Using HQPlayer and DAC8 DSD i heard some SQ differences going from PCM 16/44 to DSD 128 ... Do you estimate it is technically "only loose" ? However i agree DSD256 is quite better than DSD128 ...

 

I mean in comparison to higher DSD rates, not to PCM...

 

[volpone]

5 minutes ago, Miska said:

 

I mean in comparison to higher DSD rates, not to PCM...

 

 

OK sure .. I agree, sorry !

[tboooe]

16 hours ago, rando said:

The key item to keep in mind is the newer processors firmware doesn't let you access the full horsepower out of the gate.  They are designed to be utterly efficient during the majority of their lifetime that is not spent at 100% processor usage.  Once it figures out you mean business, that efficiency redoubles itself to do real work using programs such as HQ Player.  

Thank you Rando.  I get what you are saying about upgrading mobo and cpu being fairly close to just upgrading the cpu to a i7-7700k.  I definitely need look at this more closely.  Can you please explain what you mean by the above statement?  I am not following.

[acousticmood]

I can’t get consistent dsd playback from HQplayer and Roon through ultra Rendu into oppo205.  I had it working this morning but then I played a dvd and when I come back to it no go. The scrolling bar shows that it is playing as does the HQplayer zone. Nothing comes out though. If I switch in HQplayer to pcm then it plays but no sound with dsd.

[gdpr]

1 hour ago, acousticmood said:

I can’t get consistent dsd playback from HQplayer and Roon through ultra Rendu into oppo205.  I had it working this morning but then I played a dvd and when I come back to it no go. The scrolling bar shows that it is playing as does the HQplayer zone. Nothing comes out though. If I switch in HQplayer to pcm then it plays but no sound with dsd.

Can you try limiting to DSD256 to 44,1kHz multiple?

In HQplayer settings (outside Roon), disable autorate & set max. Rate to 11289600.

Dirk

[acousticmood]

Thankyou. I worked down on the bit rate and foundd that 48kx250 worked.

 

here am what I am running now and it’s working. I wonder why? Is it my server processor limitations?

[gdpr]

10 minutes ago, acousticmood said:

Thankyou. I worked down on the bit rate and foundd that 48kx250 worked.

 

here am what I am running now and it’s working. I wonder why? Is it my server processor limitations?

Problem is most certainly in your dac (or micro rendu). 

I personally would go for a more standard samplerate, 44.1 × 256, as I am not aware of any device running 48x250. I find it curious that Miska has foreseen this rate.

Dirk

[Miska]

Can you show what HQPlayer is actually putting out in each case that works vs doesn't work? Just a clip from HQPlayer main window "Format" part is enough. When you want to try different things it is better to stick to main window instead of changing the default values in the settings dialog which is only preferences/boundaries. If you want something specifically, then main window is the one that rules.

 

[Miska]

12 minutes ago, ddetaey said:

I personally would go for a more standard samplerate, 44.1 × 256, as I am not aware of any device running 48x250. I find it curious that Miska has foreseen this rate.

 

Some of those conversions from sampling rate to base rate multiples constantly give funny results.

 

For some custom devices there's support for typical clock frequencies like in this case 12 MHz and the double which is 24 MHz. The rate is upper limit that is picked up from the rate supported by the actual output device. These exact MHz figures are regularly available as clocks. This is no different from what for example Benchmark has been doing in their DACs.

 

So let's say there are 2.8224, 5.6448, 11.2896 and 22.5792 MHz supported (44.1 x64, 44.1 x128, 44.1 x256 and 44.1 x512). And you set limit to 48k x250. Then when selecting actual output rate HQPlayer sees that there's 22.5972M supported, but it is higher than the set limit, so it picks up the highest that is below the set limit, in this case 11.2896M. This selection is further on bounded by conversion rate capabilities of the filter. So in this case, if the DAC would advertise also 3.072, 6.144 and 12.288 MHz (48 x64, 48 x128 and 48 x256) and filter can do only simple integer ratios, then it would pick up 6.144 MHz because that is the rate that is below the set limit, advertised by the DAC as supported and also supported by the filter.

 

If you want to select something specific, in main window drop list, the actual rates advertised by the device in question are shown, and there you can select a specific rate.

 

[gdpr]

5 minutes ago, Miska said:

 

Some of those conversions from sampling rate to base rate multiples constantly give funny results.

 

For some custom devices there's support for typical clock frequencies like in this case 12 MHz and the double which is 24 MHz. The rate is upper limit that is picked up from the rate supported by the actual output device. These exact MHz figures are regularly available as clocks. This is no different from what for example Benchmark has been doing in their DACs.

 

So let's say there are 2.8224, 5.6448, 11.2896 and 22.5792 MHz supported (44.1 x64, 44.1 x128, 44.1 x256 and 44.1 x512). And you set limit to 48k x250. Then when selecting actual output rate HQPlayer sees that there's 22.5972M supported, but it is higher than the set limit, so it picks up the highest that is below the set limit, in this case 11.2896M. This selection is further on bounded by conversion rate capabilities of the filter. So in this case, if the DAC would advertise also 3.072, 6.144 and 12.288 MHz (48 x64, 48 x128 and 48 x256) and filter can do only simple integer ratios, then it would pick up 6.144 MHz because that is the rate that is below the set limit, advertised by the DAC as supported and also supported by the filter.

 

If you want to select something specific, in main window drop list, the actual rates advertised by the device in question are shown, and there you can select a specific rate.

 

Thanks Miska, very insightfull.

Dirk

[acousticmood]

30 minutes ago, Miska said:

Can you show what HQPlayer is actually putting out in each case that works vs doesn't work? Just a clip from HQPlayer main window "Format" part is enough. When you want to try different things it is better to stick to main window instead of changing the default values in the settings dialog which is only preferences/boundaries. If you want something specifically, then main window is the one that rules.

 

Options on the main pull down menu the 44.1x256 will play the 48k will not. Auto will also play 

 

[Quadman]

@acousticmood I think Jussi wants to see the main HQP screen when music is playing so he can see the format it is outputting.  In your last screen shot HQP is not playing anything.

 

@Miska  I have been helping two people via team viewer sessions to set up their playback with Win 10 pro (both 1809), current AMD based servers, HQPlayer (3.25.2), Roon and fidelizer pro.  And I am now having a strange issue, the exact same one for both.  First Both parties use Chinese dac's Gustard X20u and LKS DH004, also each party uses Chinese DDC's  Singxer su-1 with gustard and gustard u16 with LKS.  Both systems have in the past been able to successfully run HQP with roon and upsample all music to DSD512.

 

LKS system w U16 we flashed new firmware to the U16.  Then when setting up his system again for DSD512 playback I was able to get roon set up for DSD512 upsampling and successfully played music with roon upsampling to DSDS512.  Trying to get HQP set up everything looks good HQP sees U16 and its driver but on the Main screen when I select SDM as output mode the filters to the left of this do not switch to the filters for SDM I have set up on the settings page, they stay at the PCM filters and output is PCM only despite SDM being selected for output mode.  Roon can output DSD512, HQP only outputs PCM

 

The Gustard system with Singxer SU-1 we re-flashed its firmware, then when resetting up his system for DSD512 payback the exact same thing that happened above happens here as well.  I can successfully get roon set for DSD512 upsampling, and playing tracks confirms DSDS512 output.  HQP again looks good it sees the SU-1 and its driver, but when I go to select SDM as the output the filters stay as PCM and again output is only PCM.  Again Roon outputs DSD512, HQP only PCM output.

 

I know before you mentioned the driver/PC interaction is not allowing the switch to DSD, and that we should remove everything between PC and dac.  Which was tried before the re-flashing.  Do you have any idea why Roon can upsample to 512 and output DSD512 and HQP cannot, instead only outputting PCM?  I have never seen this issue before usually if it happened neither roon nor HQ could output DSD.

 

BTW I had been a very long time Poly-Sinc-XTR-MP 2s filter fan (In my system which has no issues), early this week I switched to Poly-sinc-EXT2, with ADSM7 and am just loving the sound.

 

[Miska]

53 minutes ago, Quadman said:

HQP again looks good it sees the SU-1 and its driver, but when I go to select SDM as the output the filters stay as PCM and again output is only PCM.  Again Roon outputs DSD512, HQP only PCM output.

 

That happens when the driver first says it support DSD, but when asked to actually switch to DSD it returns an error.

 

55 minutes ago, Quadman said:

Do you have any idea why Roon can upsample to 512 and output DSD512 and HQP cannot, instead only outputting PCM?

 

No idea...

 

The driver developer should test with HQPlayer and see why things go wrong...

 

[elan120]

1 hour ago, Quadman said:

@acousticmood I think Jussi wants to see the main HQP screen when music is playing so he can see the format it is outputting.  In your last screen shot HQP is not playing anything.

 

@Miska  I have been helping two people via team viewer sessions to set up their playback with Win 10 pro (both 1809), current AMD based servers, HQPlayer (3.25.2), Roon and fidelizer pro.  And I am now having a strange issue, the exact same one for both.  First Both parties use Chinese dac's Gustard X20u and LKS DH004, also each party uses Chinese DDC's  Singxer su-1 with gustard and gustard u16 with LKS.  Both systems have in the past been able to successfully run HQP with roon and upsample all music to DSD512.

 

LKS system w U16 we flashed new firmware to the U16.  Then when setting up his system again for DSD512 playback I was able to get roon set up for DSD512 upsampling and successfully played music with roon upsampling to DSDS512.  Trying to get HQP set up everything looks good HQP sees U16 and its driver but on the Main screen when I select SDM as output mode the filters to the left of this do not switch to the filters for SDM I have set up on the settings page, they stay at the PCM filters and output is PCM only despite SDM being selected for output mode.  Roon can output DSD512, HQP only outputs PCM

 

The Gustard system with Singxer SU-1 we re-flashed its firmware, then when resetting up his system for DSD512 payback the exact same thing that happened above happens here as well.  I can successfully get roon set for DSD512 upsampling, and playing tracks confirms DSDS512 output.  HQP again looks good it sees the SU-1 and its driver, but when I go to select SDM as the output the filters stay as PCM and again output is only PCM.  Again Roon outputs DSD512, HQP only PCM output.

 

I know before you mentioned the driver/PC interaction is not allowing the switch to DSD, and that we should remove everything between PC and dac.  Which was tried before the re-flashing.  Do you have any idea why Roon can upsample to 512 and output DSD512 and HQP cannot, instead only outputting PCM?  I have never seen this issue before usually if it happened neither roon nor HQ could output DSD.

 

BTW I had been a very long time Poly-Sinc-XTR-MP 2s filter fan (In my system which has no issues), early this week I switched to Poly-sinc-EXT2, with ADSM7 and am just loving the sound.

 

Interesting problem...I am not sure if I know the answer either, but since I also have SU-1 as well as AMD CPU server with Win10Pro, running HQP through Roon, I can only help confirm the firmware for SU-1 is fine in my setup running upsampling from 44.1K to DSD512 either 2's or non-2's filters, so issue might not be firmware related.