HQ Player

[Miska]

7 hours ago, jamesg11 said:

Just now updated ultraRendu to 2.8 via sd card, sonicorbiter update.

On hqp 3.latest, mac mini, no other change to anything, & rme dac is now muting with gain overload, still on dsd direct.

 

Issue? Need to now go to hqp4? (was leaving til pc upgrade).

Or is this likely a sonicorbiter sw issue?

 

If you have such issue, it is good to test without a NAA, by connecteding DAC directly to the HQPlayer server. That rules out number of network and NAA issues and helps locating whereabouts of the problem.

 

You can also just use HQPlayer v4 trial to see if it makes any difference or not. It can be installed in parallel with v3.

 

[Quadman]

4 hours ago, StreamFidelity said:

 

I set all multipliers to 41 = 4.1GHz. This permanent overclocking ensures a homogeneous, fluid sound image in my ears.

Wow surprised I thought the EC modulators were all about speed and more speed, yet at EC7 and 256 you can do most filters, Sinc L is crazy with number of taps, probably more intensive then XTR filters even.  I can't do dsd128 with sinc L (non EC) with my 3GB 1060 GPU and Ryzen 1800X OC'd to 4.0Ghz, yet I can do DSD512 with sinc S (NonEC of course).

[Yviena]

The new Sinc-L at 768khz LNS15 sounded excellent  with my HD800 till I got to a track that sounded complete shit with it, the track sounded closed-in in soundstage, and just bad overall, for some reason ext2  fixes the closed-in soundstage somewhat on that track, maybe that song really needed the apodizing part, but it sounds kind of good/much better when sent to the shitty built in speakers on my OLED TV compared when going through my HD800 so I dunno why it sounds so bad with the headphones without ext2 filter.

 

 

[Rune]

Just tried the following on my i9-9900KS without GPU running embedded.

sinc-L, ASDM7EC, DSD256  dropouts

sinc-L, ASDM7, DSD256  dropouts

sinc-L, ASDM7EC, DSD128  works

 

Desktop with Ryzen 3600X with GTX 1080 TI

sinc-L, ASDM7EC, DSD256  dropouts

sinc-L, ASDM7, DSD256  works

 

So my conclusion for sinc-L DSD256 is that you need a GPU.

[pdg540]

2 hours ago, Rune said:

Just tried the following on my i9-9900KS without GPU running embedded.

sinc-L, ASDM7EC, DSD256  dropouts

sinc-L, ASDM7, DSD256  dropouts

sinc-L, ASDM7EC, DSD128  works

 

Desktop with Ryzen 3600X with GTX 1080 TI

sinc-L, ASDM7EC, DSD256  dropouts

sinc-L, ASDM7, DSD256  works

 

So my conclusion for sinc-L DSD256 is that you need a GPU.

 

My 9900k at 5GHZ and no GPU also stutters with non-EC and sinc L. From your results, it's looking like the 7EC with sinc-L (2ch, DSD256) is going to take more than a 1080Ti with a 9900k. Is your 9900ks overclocked and does your 1080Ti have 11gb ram? Do you think the 1080Ti with the 9900 would be any different than with the Ryzen or are the 2 CPU's pretty equal?

[Miska]

4 hours ago, Rune said:

So my conclusion for sinc-L DSD256 is that you need a GPU.

 

That has been my conclusion too...

 

1 hour ago, pdg540 said:

Do you think the 1080Ti with the 9900 would be any different than with the Ryzen or are the 2 CPU's pretty equal?

 

If 1080Ti works with Ryzen, it should certainly work with 9900K too. When using GPU offload, 9900K should work equally well to 9900KS, because less cores are needed to Turbo Boost can kick few cores to 5 GHz on 9900K as well. No need for all-core 5G Turbo. So sinc-L with ASDM7EC to DSD256 should work. Given that sinc-L at 256x works on that GPU.

 

[Miska]

13 hours ago, Yviena said:

The new Sinc-L at 768khz LNS15 sounded excellent  with my HD800 till I got to a track that sounded complete shit with it, the track sounded closed-in in soundstage, and just bad overall, for some reason ext2  fixes the closed-in soundstage somewhat on that track, maybe that song really needed the apodizing part, but it sounds kind of good/much better when sent to the shitty built in speakers on my OLED TV compared when going through my HD800 so I dunno why it sounds so bad with the headphones without ext2 filter.

 

This is typical for non-apodizing filters. If the track is technically good, it works with non-apodizing filter. If it has technical issues, those issues are faithfully reproduced with all the detail. While apodizing filters like ext2 or sinc-S give you pretty consistent performance across the board, even if the track has certain type of technical issues. Of course nothing will fix a truly bad track, so apodizing filters only fix limited typical technical categories of issues.

 

I have three filters at the moment depending on material I use. sinc-S when I feel I need some extra edge, ext2 by default for most things, and for good old prog rock tracks and such short-mp.

 

[MemoryPlayer]

Can be the same new version license of the HQPlayer Desktop be used for Mac, Windows and Linus?

 

I'm testing the 4.6.0 trial with Unbutu Studio and the sound is better than Mac and Windows!

My DAC goes to DSD512, but with Windows 10 Pro new releases + Asio and Mac OS go only to DSD256!

 

I'm comparing the three versions with Parallels and now I understand why Miska's preference is Linux!

[Yviena]

6 hours ago, Miska said:

 

This is typical for non-apodizing filters. If the track is technically good, it works with non-apodizing filter. If it has technical issues, those issues are faithfully reproduced with all the detail. While apodizing filters like ext2 or sinc-S give you pretty consistent performance across the board, even if the track has certain type of technical issues. Of course nothing will fix a truly bad track, so apodizing filters only fix limited typical technical categories of issues.

 

I have three filters at the moment depending on material I use. sinc-S when I feel I need some extra edge, ext2 by default for most things, and for good old prog rock tracks and such short-mp.

 

How feasible would it be to create something that analyzes the track and notifies us if it needs apodizing or not?

 

That track i mentioned must have something seriously  wrong as even some remixes of a heavy metal band i found on youtube/soundcloud by a popular remixer with a flac download sounds much better no need for apodizing there, even if soundcloud remixes aren't usually known for good mastering.

[pdg540]

7 hours ago, Miska said:

 

That has been my conclusion too...

 

 

If 1080Ti works with Ryzen, it should certainly work with 9900K too. When using GPU offload, 9900K should work equally well to 9900KS, because less cores are needed to Turbo Boost can kick few cores to 5 GHz on 9900K as well. No need for all-core 5G Turbo. So sinc-L with ASDM7EC to DSD256 should work. Given that sinc-L at 256x works on that GPU.

 

 

1080Ti hasn't been confirmed with any CPU to run 7EC/sinc-L/256DSD as far as I've understood. Only been confirmed not to work with Ryzen+1080Ti.

[Miska]

14 hours ago, pdg540 said:

1080Ti hasn't been confirmed with any CPU to run 7EC/sinc-L/256DSD as far as I've understood. Only been confirmed not to work with Ryzen+1080Ti.

 

If it works with sinc-L using any modulator to DSD256, then it must be fine, because filter runs on the GPU and the modulators never run on a GPU. Modulator capabilities is up to CPU.

 

[Miska]

17 hours ago, MemoryPlayer said:

Can be the same new version license of the HQPlayer Desktop be used for Mac, Windows and Linus?

 

HQPlayer 4 Desktop license is OS agnostic, so you can freely choose any of the three OS.

 

[Miska]

14 hours ago, Yviena said:

How feasible would it be to create something that analyzes the track and notifies us if it needs apodizing or not?

 

I would rather make it do the work during playback, rather than doing it during content scan. Because analyzing the content during scan would make the scan take much much longer. Not sure how much of an issue it is, but still. If you would analyze terabyte of music from NAS, HQPlayer would need to fetch terabyte amount of data which already by itself takes quite a while. During playback it is not as painful, but you'd only know once you play.

 

14 hours ago, Yviena said:

That track i mentioned must have something seriously  wrong as even some remixes of a heavy metal band i found on youtube/soundcloud by a popular remixer with a flac download sounds much better no need for apodizing there, even if soundcloud remixes aren't usually known for good mastering.

 

Steven Wilson? I don't know how RedBook versions of his remixes look like (depends how it was converted from hires), but his hires ones I've seen are certainly good and don't need apodizing filter.

 

[jamesg11]

17 hours ago, Yviena said:

How feasible would it be to create something that analyzes the track and notifies us if it needs apodizing or not?

 

That track i mentioned must have something seriously  wrong as even some remixes of a heavy metal band i found on youtube/soundcloud by a popular remixer with a flac download sounds much better no need for apodizing there, even if soundcloud remixes aren't usually known for good mastering.

What are the “technical issues” in the music referred to? DR? Has this been covered somewhere or can someone elaborate?

[DancingSea]

On 6/13/2020 at 11:21 AM, Miska said:

I have three filters at the moment depending on material I use. sinc-S when I feel I need some extra edge, ext2 by default for most things, and for good old prog rock tracks and such short-mp.

 

This is very helpful.  In general (or specifically if you're brave) what sorts of filters/ modulators) provide a more smooth, relaxed sound?  I'm less interested in resolution.  With countless combinations (thousands?), a nudge in the right direction would be helpful.  I'm adding short-mp to that list, any others that take the edge off?

[jamesg11]

On 6/13/2020 at 5:32 AM, Miska said:

 

If you have such issue, it is good to test without a NAA, by connecteding DAC directly to the HQPlayer server. That rules out number of network and NAA issues and helps locating whereabouts of the problem.

 

You can also just use HQPlayer v4 trial to see if it makes any difference or not. It can be installed in parallel with v3.

 

Sorted. Now hqp 3.25.6 on high sierra. End of road for for 3!

 

Just to check that this different presentment of info means my rme dac is still outputting dsd256 in pure dsd setting - the dac always showed 768? Or has it somehow flipped to pcm ...?

 

[AnotherSpin]

3 hours ago, DancingSea said:

 

This is very helpful.  In general (or specifically if you're brave) what sorts of filters/ modulators) provide a more smooth, relaxed sound?  I'm less interested in resolution.  With countless combinations (thousands?), a nudge in the right direction would be helpful.  I'm adding short-mp to that list, any others that take the edge off?

 

I use sinc-L and Gauss1 for a very smooth and natural sound in PCM with R2R DAC. For old rock albums (Led Zeppelin, etc.) apodizing sinc-S (+ Gauss1) would fix certain issues.

[Miska]

11 hours ago, jamesg11 said:

What are the “technical issues” in the music referred to? DR? Has this been covered somewhere or can someone elaborate?

 

Digital decimation filters used in most of even remotely modern ADCs, and also some mastering tools used to make rate conversions create set of problems. Luckily some of these can be fixed.

 

[Miska]

3 hours ago, jamesg11 said:

Sorted. Now hqp 3.25.6 on high sierra. End of road for for 3!

 

Just to check that this different presentment of info means my rme dac is still outputting dsd256 in pure dsd setting - the dac always showed 768? Or has it somehow flipped to pcm ...?

 

 

Based on this screenshot, output is 768k PCM, converted from DSD256 file.

 

ADI-2 doesn't indicate DSD inputs, so the DAC will always just state the relevant PCM rate (DSD is always through DoP to it).

 

[Miska]

3 hours ago, DancingSea said:

This is very helpful.  In general (or specifically if you're brave) what sorts of filters/ modulators) provide a more smooth, relaxed sound?  I'm less interested in resolution.  With countless combinations (thousands?), a nudge in the right direction would be helpful.  I'm adding short-mp to that list, any others that take the edge off?

 

I would say you get pretty nice sound with poly-sinc-short-mp (or -lp if you like that more) and ASDM5(EC) modulator. Use the EC modulator if you have enough CPU power for it, otherwise the non-EC. This doesn't shave off details, but gives relaxed sound.

 

Surprisingly poly-sinc-ext2 instead as a filter may provide some similarity although being technically very different.

 

This gives you only few things to try. But I don't have your ears and system, so over time you could experiment with other things as well.

 

[jamesg11]

2 hours ago, Miska said:

 

Based on this screenshot, output is 768k PCM, converted from DSD256 file.

 

ADI-2 doesn't indicate DSD inputs, so the DAC will always just state the relevant PCM rate (DSD is always through DoP to it).

 

Thanks, in preferences how do I ensure that this is always an upsample to dsf256, regardless of input file? Previously, had this ... now forgotten how to set it.

[Miska]

3 hours ago, jamesg11 said:

Thanks, in preferences how do I ensure that this is always an upsample to dsf256, regardless of input file? Previously, had this ... now forgotten how to set it.

 

In DSDIFF/DSF Settings check that Direct SDM is disabled. Then in main window, select "SDM (DSD)" from the right-most drop list, and from rate list right next to it, select 44.1k x256.

[MemoryPlayer]

On 4/26/2019 at 7:54 PM, Miska said:

 

Price is now the same for both HQPlayer Embedded and HQPlayer Desktop. 199 EUR net, while previous version has been running 119 EUR net for years. But now you get all three platforms covered at once.

 

Unfortunately there's just a limit how far single person can stretch and I would love to be able to hire a student to help out on some of the things. That would enable more things to happen at faster pace.

 

Sorry for my previous question Miska...

Checking back I saw now we get all three platforms (Mac, Linux and Windows) with only one license!

[shahed99]

My server has Ryzen 3950X (16 core, 32 threads). Yesterday, I installed ProcessLasso and assign CPU affinity for HQPlayer and Roon. The sound improvement was immediate. I did this six years ago on my Intel 4770T and the improvement wasn't as pronounced as it is now.

 

@Miska Can you suggest how many cores/threads I should assign to HQP? Should I also turn-off SMT for HQP? When I turn off SMT for HQP in Process Lasso, only even numbered core/threads are selected (0, 2, 4, 6, ..); probably those are on the same physical node? Would HQP also benefit from increasing priority to real-time?

[DancingSea]

13 hours ago, Miska said:

 

I would say you get pretty nice sound with poly-sinc-short-mp (or -lp if you like that more) and ASDM5(EC) modulator. Use the EC modulator if you have enough CPU power for it, otherwise the non-EC. This doesn't shave off details, but gives relaxed sound.

 

Surprisingly poly-sinc-ext2 instead as a filter may provide some similarity although being technically very different.

 

This gives you only few things to try. But I don't have your ears and system, so over time you could experiment with other things as well.

 

 

Thanks for the suggestions.  My Mac Mini does all the EC modulators to DSD 128 (the max of my DirectStream Junior Dac) with no struggle.  I’ve tried ext2 in the past, but always seem to prefer the long filters.  Though last night I listened to Kenny Burrell’s “Midnight Blue” with ext2/ ASDM5(EC) and it was the best I’ve ever heard that favorite recording.

 

Is “MP” inherently more relaxed than “LP”?