HQ Player

[Miska]

On 12/22/2019 at 5:04 AM, pcmchild said:

 

Not many people use chord dacs around here. I do, I use TT2, but don't have mscaler. Here a snapshot of my settings. Surrounded by red line the ones you should set.

Some people also like Sinc-M filter, but poly-sinc-long-lp I read should be the one more similar to mscaler?

For output you can use both WASAPI or ASIO.

 

If you are using USB connection, like in this case, you can leave "DAC Bits" to "Default" practically meaning 32-bit output.

 

You should always use dither, because otherwise the needed reduction of resolution at the output ends up with distortion... In this kind of case, TPDF dither is good starting point.

 

 

[Miska]

On 12/21/2019 at 2:36 PM, ronnie.sengupta said:

When I try 44.1 PCM to poly-sinc-xtr DSD 512, I get a lot of dropouts on my I7 - 7700k + GTX 1080 PC. The CPU runs at about 62% load without Cuda off load and with Cude off load, I get 25% CPU usage and 80% GPU usage.

 

If only 62% of the CPU is used, I was wondering if the bottleneck is with RAM speed. I am currently using DDR 4 2400Mhz RAM. I am on Windows 10 home edition running HQP 3.25.6

 

The xtr filter at DSD256 runs without any problem and sounds wonderful.

 

Since 7700K has HyperThreading, half of the cores shown by by the OS are virtual "threads" of the CPU. These cannot do any real work, they just help on being more efficient on context switching between multiple tasks on the CPU. So when you reach 50% load as indicated by the OS, you are already 100% load on all the real CPU cores. So if you manage to get 62%, it means you have exceeded the real computing capacity. So when you look at load figures with threaded CPUs, any figures between 50 and 100% are pretty bogus.

 

The case you are describing (assuming target of 44.1k x512, not 48k x512) is about on the edge of capabilities of that hardware. With some careful tuning it may work. 44.1k to 48k x512 won't work with any hardware I'm aware of.

 

[Hensema]

19 hours ago, Miska said:

You should always use dither, because otherwise the needed reduction of resolution at the output ends up with distortion... In this kind of case, TPDF dither is good starting point.

is dithering needed when 32 bit is selected? I can understand why it’s beneficial when 16 bit is selected. The noise of the dac itself will “dither” on that high bitrates?

[Miska]

18 hours ago, Hensema said:

is dithering needed when 32 bit is selected? I can understand why it’s beneficial when 16 bit is selected. The noise of the dac itself will “dither” on that high bitrates?

 

It is still needed to get undistorted data to the DAC's DSP. There's no harm doing dither to 32-bit, but there is harm not doing it.

 

[StreamFidelity]

It's worth checking the Signalyst site regularly:

 

Dec 19 2019	HQPlayer 4 Desktop 4.2.3 released. Small CPU core allocation tunings and fixes.
Dec 17 2019	HQPlayer 4 Desktop 4.2.2 released. Component updates and bug fixes. Also includes further fine tuning of core allocations.

Thank you Miska for the regular improvements. The cores are loaded more evenly in my system than before. A very great sound. 👍

 

 

[Outlaw]

Looking at BIOS option for CPU core there is a option to sync all cores.Is there any advantage to this for hqplayer so all cores run at same speed.

[ronnie.sengupta]

I managed to get 44.1 PCM -> DSD 256 working with ADM7EC + poly-sinc-xtr-lp running on my i7-7700k

 

Overclocked to 4.7ghz with hyper threading disabled. CPU running at 98% with CPU temps around 69 C

 

Now, I need to find a reasonably priced CPU which will run the above settings at 50% load!

 

[Miska]

2 hours ago, ronnie.sengupta said:

I managed to get 44.1 PCM -> DSD 256 working with ADM7EC + poly-sinc-xtr-lp running on my i7-7700k

 

Overclocked to 4.7ghz with hyper threading disabled. CPU running at 98% with CPU temps around 69 C

 

Now, I need to find a reasonably priced CPU which will run the above settings at 50% load!

 

 

Good achievement to get load to 100% without dropouts! Not really much spare left for the OS. You could keep HyperThreading enabled to get a little bit better performance. And the total load figure drops to half because there are twice as many logical CPUs for the same number of physical cores, not that it would change anything from practical point of view.

 

i9-9900K(S) is probably what you are looking for. Pretty much same clock frequencies but twice as many cores.

 

Another alternative is to use some RTX-series GPU, like 2080 to run the filters. But it doesn't really save much money. And 9th gen i9 is still somewhat faster than 7th gen i7, clock-to-clock.

 

[Miska]

22 hours ago, Outlaw said:

Looking at BIOS option for CPU core there is a option to sync all cores.Is there any advantage to this for hqplayer so all cores run at same speed.

 

In some cases it can be beneficial. But that usually limits achievable clock rate, so it may have negative impact that way.

 

HQPlayer load is rarely fully even between all cores, so in such cases it is beneficial to have high load cores running at higher turbo speeds and the ones with lighter load at lower clocks. Keeps temperatures down and helps getting higher boost clocks on the high loaded ones.

 

[hykhleif]

On 12/22/2019 at 5:04 AM, pcmchild said:

 

Not many people use chord dacs around here. I do, I use TT2, but don't have mscaler. Here a snapshot of my settings. Surrounded by red line the ones you should set.

Some people also like Sinc-M filter, but poly-sinc-long-lp I read should be the one more similar to mscaler?

For output you can use both WASAPI or ASIO.

thanks i added dither tpdf and that nailed it when using ur settings

[hykhleif]

On 12/25/2019 at 12:25 PM, Miska said:

 

It is still needed to get undistorted data to the DAC's DSP. There's no harm doing dither to 32-bit, but there is harm not doing it.

 

yes using dither makes it sounding better on tt2

 

can you tell me the difference between poly sing long lp and sinc m

[Miska]

5 minutes ago, hykhleif said:

can you tell me the difference between poly sing long lp and sinc m

 

There is very little in common between the two. Both are apodizing and have steep roll-off.

 

[hykhleif]

2 minutes ago, Miska said:

 

There is very little in common between the two. Both are apodizing and have steep roll-off.

 

when u say little in common u mean they are very different and can you say what both differ in sound presentation, like is one more warmer sounding than the other

 

also do u recommend to set minimum volume to -20 like in the settings above

[jimdukey]

I don't understand why Vol settings can be so low.

I use either -6 and 0, or -3 and -3.

I tried a very low setting and got no sound at all.

So what's the deal?

[hykhleif]

4 minutes ago, jimdukey said:

I don't understand why Vol settings can be so low.

I use either -6 and 0, or -3 and -3.

I tried a very low setting and got no sound at all.

So what's the deal?

i am asking bout minimum volume not max volume

[jimdukey]

Me too.

[Miska]

2 minutes ago, hykhleif said:

when u say little in common u mean they are very different and can you say what both differ in sound presentation, like is one more warmer sounding than the other

 

also do u recommend to set minimum volume to -20 like in the settings above

 

I'm not trying to describe much in subjective terms, because it is always very personal experience. poly-sinc-ext2 is likely sonically closer to sinc-M than poly-sinc-long is. There is likely not so much difference in "warmth" between these though.

 

[Miska]

Min and Max volume settings just set range for the volume control (knob in the main window that is also reflected to HQPlayer Client, Roon, etc). You then need to set the actual volume to something suitable. Of course if min and max are the same, the range is span is 0 dB's so then there's not much to adjust...

 

[jimdukey]

But Why choose a very low setting in the first place?

I can't tell any difference between -3 and -3, or -6 and 0.

I would think it's more important to turn down the Max Level, to prevent clipping.

What does -20 or even lower for Minimum Vol accomplish?

And how does a Min setting of -6 affect Max Level/clipping when Max is set to 0? 

Please bear in mind I don't understand even half of what is discussed here...!!!

Love HQ, tho.

[Miska]

18 minutes ago, jimdukey said:

But Why choose a very low setting in the first place?

I can't tell any difference between -3 and -3, or -6 and 0.

I would think it's more important to turn down the Max Level, to prevent clipping.

What does -20 or even lower for Minimum Vol accomplish?

And how does a Min setting of -6 affect Max Level/clipping when Max is set to 0? 

Please bear in mind I don't understand even half of what is discussed here...!!!

 

If you have a DAC connected directly to your power amp, turning volume up to -3 would probably both make the power amp clip and also fry the loudspeakers. There a setting of -26 or something would be quite typical for normal listening. For me, even volume setting between non-compressed classical music and very compressed pop differs by more than 10 dB. For quiet night time background music I can turn volume down to even -50 dB.

 

For my headphone listening typical volume settings I have can be around -20 dB or so.

 

For measurements, I have minimum set to -120 dB and maximum +6. But it becomes a bit difficult to adjust precisely.

 

Default minimum is -60 dB and maximum 0 dB and thus the range spans 60 dB which is quite useful for normal volume control use.

[Miska]

If one wants to fine-tune performance of a system with pre-amp or integrated one way is to do following:

 

1) Turn down volume from amp

2) Set HQPlayer volume to -3 dBFS

3) Turn volume control on amp as high as you would ever use for normal listening

4) Turn down volume from HQPlayer to a level that you want to listen at the moment

 

In many cases this gives you best SNR/THD+N figures from the amplifier and DAC. And your volume control is accessible together with the music playback from the same place. And at least it is easy to try.

 

[Jerry]

5 hours ago, Miska said:

If one wants to fine-tune performance of a system with pre-amp or integrated one way is to do following:

 

1) Turn down volume from amp

2) Set HQPlayer volume to -3 dBFS

3) Turn volume control on amp as high as you would ever use for normal listening

4) Turn down volume from HQPlayer to a level that you want to listen at the moment

 

In many cases this gives you best SNR/THD+N figures from the amplifier and DAC. And your volume control is accessible together with the music playback from the same place. And at least it is easy to try.

 

 

Dear @Miska, since ESS9028Pro (or my DAP iBasso DX220) use software volume control, to get the best SNR/THD+N, I should set the DAP to maximum gain, maximum volume (putting the DAP in DAC mode), and control the actual music volume with HQplayer's volume control? Could you please also explain how the software volume control in HQPlayer is better than the DAC's implementation? As always, appreciate your great work!

 

 

 

[ronnie.sengupta]

9 hours ago, Miska said:

 

Good achievement to get load to 100% without dropouts! Not really much spare left for the OS. You could keep HyperThreading enabled to get a little bit better performance. And the total load figure drops to half because there are twice as many logical CPUs for the same number of physical cores, not that it would change anything from practical point of view.

 

i9-9900K(S) is probably what you are looking for. Pretty much same clock frequencies but twice as many cores.

 

Another alternative is to use some RTX-series GPU, like 2080 to run the filters. But it doesn't really save much money. And 9th gen i9 is still somewhat faster than 7th gen i7, clock-to-clock.

 

 

With Hyper threading enabled, the CPU usage comes down to around 80% and there is constant stuttering, which is a bit strange.

 

I was a bit hasty in posting my earlier results, with Hyper threading disabled, I do get some stutter, but once or twice in a song and sometimes no stutter at all.

 

i9 9900KS is not yet available in India. I could try upgrading from  my GTX 1080 to 2080, or wait till the processor becomes available in India.

 

[Jerry]

6 minutes ago, ronnie.sengupta said:

 

With Hyper threading enabled, the CPU usage comes down to around 80% and there is constant stuttering, which is a bit strange.

 

I was a bit hasty in posting my earlier results, with Hyper threading disabled, I do get some stutter, but once or twice in a song and sometimes no stutter at all.

 

i9 9900KS is not yet available in India. I could try upgrading from  my GTX 1080 to 2080, or wait till the processor becomes available in India.

 

 

Just to reiterate, with proper cooling, a 6700K overclocked to 4.7G and a GTX1080 are enough to achieve EC variant DSD256 with the most demanding non-2s xtr filter for red book materials. Try to play around the HQPlayer settings and you will get there...

[shang_2000]

@Miskacan allo usbridge signature use naa 3601 for pi3b?