Jump to content
IGNORED

HQ Player


Recommended Posts

I agree, 3.15.1 at DSD512 sounds great here. I also reinstalled 3.15, but it didn't stay in very long.

 

I went from 3.14.4 to 3.15.1 so don't have 3.15.0 to try! I have noticed some weird CPU behavior now but don't know how far back it goes. Sometimes HQP will jump 3-5 seconds at a time, or just run some seconds quickly then wait then run some seconds quickly, again. I can stop/start play and get exact same result or it will change to normal speed running. Music speed is not affected, but maybe the sound quality is?

 

Weird HQP CPU.PNG

 

You can see the sinusoidal CPU behavior. It is hard to see unless you change Task Manager Performance graph Update Speed to Fast (default Normal smooths the response too much). If you split it to show all 8 cores you can see that CPU Utilization hits 100% sometimes for some cores. I don't think this can be good. So, watch out about CPU Utilization if you get dropouts. Maybe some cores are hitting 100%. Note, my music flow has not been affected (no dropouts in any PCM to DSD or DSD to DSD upsampling).

Link to comment
You can see the sinusoidal CPU behavior. It is hard to see unless you change Task Manager Performance graph Update Speed to Fast (default Normal smooths the response too much). If you split it to show all 8 cores you can see that CPU Utilization hits 100% sometimes for some cores. I don't think this can be good. So, watch out about CPU Utilization if you get dropouts. Maybe some cores are hitting 100%. Note, my music flow has not been affected (no dropouts in any PCM to DSD or DSD to DSD upsampling).

 

Do you have CUDA offload enabled?

Link to comment
I observed this unusual 3.15.1 behavior without CUDA on i5.

 

... and I checked that also on 3.15.0, see the picture

2017-03-13 19_25_23-Task Manager.png

 

It is for 96k to 6.1M poly-sinc-xtr-2s / ASDM7 with matrix pipeline & impulse response WAVs for stereo to binaural processing on i5 4300M turbo boosted to 3MHz without CUDA. I like the sound of xtr filters on 3.15.0, so I don't complain if it should be so. :)

 

My only trouble with 3.15.0 is that poly-sinc-xtr[-mp]-2s initializes within few seconds from 44.1k to 5.6M but it takes endless time from 44.1k to 6.1M (I needed to kill HQPlayer). So I have to use the Auto rate family setting.

i7 11850H + RTX A2000 Win11 HQPlayer ► Topping HS02 ► 2x iFi iSilencer ► SMSL D300 ► DIY headamp DHA1 ► HiFiMan HE-500
Link to comment

I have to agree... 3.15.0 sounds different to 3.15.1. The newer version has something brittle about it.

 

Poly-sinc short/ASDM7 44.1 kHz —> 5.6MHz.

Roon client on iPad/MacBookPro

Roon Server & HQPlayer on Mac Mini 2.0 GHz i7 with JS-2

LPS-1 & ultraRendu → Lampizator Atlantic → Bent Audio TAP-X → Atma-sphere M60 → Zero autoformers → Harbeth Compact 7 ES-3

Link to comment
I went from 3.14.4 to 3.15.1 so don't have 3.15.0 to try! I have noticed some weird CPU behavior now but don't know how far back it goes. Sometimes HQP will jump 3-5 seconds at a time, or just run some seconds quickly then wait then run some seconds quickly, again. I can stop/start play and get exact same result or it will change to normal speed running. Music speed is not affected, but maybe the sound quality is?

 

[ATTACH=CONFIG]33997[/ATTACH]

 

You can see the sinusoidal CPU behavior. It is hard to see unless you change Task Manager Performance graph Update Speed to Fast (default Normal smooths the response too much). If you split it to show all 8 cores you can see that CPU Utilization hits 100% sometimes for some cores. I don't think this can be good. So, watch out about CPU Utilization if you get dropouts. Maybe some cores are hitting 100%. Note, my music flow has not been affected (no dropouts in any PCM to DSD or DSD to DSD upsampling).

 

That may also be sampling artifact because there is no clear anti-aliasing and reconstruction filter. So CPU load is calculated as utilization percentage over certain sampling interval, while the the load is always spiky (run-and-wait). Depending on how load spikes happen to go in sync with the CPU load sampling window.

 

IOW, the load is always 100% on the utilized cores when processing is being done. 50% load means half of the measurement interval time CPU was doing work and half of the time it was waiting. 25% load means 1/4th of the time CPU was doing work and 3/4ths of the time it was waiting. Most CPU load graph plotting utilities give some possibility to adjust the time window that is considered for calculating the utilization percentage. For example for 10 second window, 50% load means CPU was doing work for 5 seconds and waiting for 5 seconds (running do-nothing-loop). This could mean 5 times 1 second run sprints or 500 times 10 millisecond run sprints.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
Miska - CUDA support multiple graphics cards in one system, does HQP Cuda offload support multiple GPU-s?

 

Not directly, unless CUDA decides to give different GPUs for different instances. So HQPlayer just asks CUDA that "give me a GPU that matches my requirements" and then uses it. I could add more explicit multi-GPU support, but I have not found straightforward way to decide between different GPUs if those happen to be different models.

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment

Miska, is NVIDA Quadro series good alternative for GeForce cards when used with HQPlayer?

For example, I'd add Quadro P2000 (maybe P4000) to i7-7000 for upsampling to DSD512 and XTR-MP filter, ADM7.

Vinnie Rossi LIO (AVC/Tubestage, AMP Module with built in HPF 100Hz 24dB/octave, DAC 2.0), Harbeth P3ESR, Rythmik F8

Win10 i7-7700 -> Roon -> HQPlayer DSD512- > LIO 100Hz HPF -> Harbeth P3ESR

                                                                                ->LIO  -> miniDSP <100Hz -> Rythmik F8  

 

 

 

Link to comment

By the way I just bought an iFi iDAC2, and interestingly enough it does not do the mentioned loud popping noise, it only has some really minor and quiet popping noise, which is strange as I know the iDAC2 uses the same xmos interface as the iDSD.

Link to comment

@ferenc, it's not only about using XMOS chip as such. To remove pops (which can occur for example during switching between PCM and DSD modes) many DACs implement attenuation. Some DACs use relays to attenuate signal during short time of format switch. Functionality of this attenuation is programmed in firmware. So FW upgrade may help with the issue.

 

Other known type of solution: Some software players play pause during playback start or during PCM <-> DSD switch to avoid pops.

i7 11850H + RTX A2000 Win11 HQPlayer ► Topping HS02 ► 2x iFi iSilencer ► SMSL D300 ► DIY headamp DHA1 ► HiFiMan HE-500
Link to comment

Miska:

 

My W4S DAC2 DSDse is being returned to me tomorrow after being upgraded to the new Sabre ES9038Pro chip. The chip apparently has new filters as follows:

 

Slow rolloff, linear phase / Slow rolloff, minimum phase / Fast rolloff, linear phase / Fast rolloff, minimum phase / Apodizing fast rolloff, linear phase(default) / Hybrid fast rolloff, minimum phase / Brickwall

 

My question is:

 

How do these filters integrate/coordinate with your filters? Are there certain combinations to try together or to avoid? Or does it not matter and I just have more combinations to try to find what I like? I'm currently liking poly-sinc-shrt-mp-xtr/ADSM7.

 

Also, similar question as there are now some adjustable Bandwidth/IIR settings, <50k (default), 50k, 60k, 70k?

 

I mostly use HQP to transcode from Redbook to DSD128 and now hopefully DSD256?

 

Thanks!

Bob

 

Mac Mini M1 12 volt dc > Roon > HQ Player to DSD 256 > Fibre to EtherRegen w/LPS1.2 and BG7TBL OCXO > Sonore microRendu v1.3 > IsoRegen > Denafrips Iris > i2s  > Denafrips Pontus II > Schiit Freya+ w/ Linlai E-6SN7's >  Nord One Up NCore 500 monoblocks REV D w/SI990Enh op amps > Martin Logan Impression 11A w/ dual Rythmik E15HP2 subs. Supra Cat8, JPS Labs Superconductor+ cables

 

 

Link to comment
I mostly use HQP to transcode from Redbook to DSD128 and now hopefully DSD256?

Then the Sabre PCM upsampling filters you mentioned are not in use. The point of using HQPlayer is to use HQPlayer filters instead of those in your DAC chip.

 

In principle:

 

a) You want to use HQPlayer because HQPlayer filters running on powerful PC are on higher qualitative level than those in the resource constrained DAC chip. If your DAC is DSD capable, you typically want to upsample to max. DSD rate your DAC supports to bypass DAC filtering (possible with Sabre) and in some cases also delta-sigma modulation (not the Sabre case).

 

or

 

b) You want to use filters of your DAC but then you don't need HQPlayer.

i7 11850H + RTX A2000 Win11 HQPlayer ► Topping HS02 ► 2x iFi iSilencer ► SMSL D300 ► DIY headamp DHA1 ► HiFiMan HE-500
Link to comment
Then the Sabre PCM upsampling filters you mentioned are not in use. The point of using HQPlayer is to use HQPlayer filters instead of those in your DAC chip.

 

In principle:

 

a) You want to use HQPlayer because HQPlayer filters running on powerful PC are on higher qualitative level than those in the resource constrained DAC chip. If your DAC is DSD capable, you typically want to upsample to max. DSD rate your DAC supports to bypass DAC filtering (possible with Sabre) and in some cases also delta-sigma modulation (not the Sabre case).

 

or

 

b) You want to use filters of your DAC but then you don't need HQPlayer.

 

Understood the principle and makes perfect sense if the chip automatically bypasses it's internal filters when receiving DSD which I wasn't sure was happening since the dac doesn't have any transcoding/upsampling feature itself or any way to defeat the filters manually.

Bob

 

Mac Mini M1 12 volt dc > Roon > HQ Player to DSD 256 > Fibre to EtherRegen w/LPS1.2 and BG7TBL OCXO > Sonore microRendu v1.3 > IsoRegen > Denafrips Iris > i2s  > Denafrips Pontus II > Schiit Freya+ w/ Linlai E-6SN7's >  Nord One Up NCore 500 monoblocks REV D w/SI990Enh op amps > Martin Logan Impression 11A w/ dual Rythmik E15HP2 subs. Supra Cat8, JPS Labs Superconductor+ cables

 

 

Link to comment
Understood the principle and makes perfect sense if the chip automatically bypasses it's internal filters when receiving DSD which I wasn't sure was happening since the dac doesn't have any transcoding/upsampling feature itself or any way to defeat the filters manually.

 

It bypasses internal filters at least to some extent, but AFAIU, still upsamples (non-bypassably) from there to some very high sample rate.

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment
It bypasses internal filters at least to some extent, but AFAIU, still upsamples (non-bypassably) from there to some very high sample rate.

 

You mean the 2nd upsampling stage where user has no choices. The user selectable PCM upsampling choices rah50 mentioned are AFAIK bypassed by feeding DSD signal to DAC.

i7 11850H + RTX A2000 Win11 HQPlayer ► Topping HS02 ► 2x iFi iSilencer ► SMSL D300 ► DIY headamp DHA1 ► HiFiMan HE-500
Link to comment
You mean the 2nd upsampling stage where user has no choices. The user selectable PCM upsampling choices rah50 mentioned are AFAIK bypassed by feeding DSD signal to DAC.

 

Yep, exactly.

 

 

Sent from my iPhone using Computer Audiophile

One never knows, do one? - Fats Waller

The fairest thing we can experience is the mysterious. It is the fundamental emotion which stands at the cradle of true art and true science. - Einstein

Computer, Audirvana -> optical Ethernet to Fitlet3 -> Fibbr Alpha Optical USB -> iFi NEO iDSD DAC -> Apollon Audio 1ET400A Mini (Purifi based) -> Vandersteen 3A Signature.

Link to comment

please check the website for details

 

works and sounds so great that I did not make any comparison, just listened

 

Just one strange thing : I'm pretty sure I checked the Multicore DSP option to start with (and I first listened to various PCM through psxtr2s) but came back to settings when I saw that some cores were hitting the 100% ceiling (nothing audible though) with DSD source. Indeed it was unchecked. Checking it made the global load raise to 44 from 35 (DSD 64 to 128, convolution). Had HQP decided to grey that option when playing DSD ?

 

Miska, do you recommend a lower load but with 100% hits or a glbal higher load with more even distribution to the cores?

What are the changes in 3.16 ? Thanks
Link to comment

Miska,

 

you stayed away from the digital vinyl temporal domain debate

 

http://www.computeraudiophile.com/f3-article-comments/article-digital-vinyl-temporal-domain-31816/index12.html#post646493

 

I use old active studio monitors that deliver 20/20K and are top guns from vinyl/experimental PCM age and per Archimago know that my TEAC 501 I feed DSD 128 and can be 384K fed as well starts to roll off at 20K : I'm not in claiming I hear ultrasonics.

 

I would greatly like your vision on my last question in this thread :

 

Is it really the same to filter 20/20K at speakers level, or DAC level, and to not sample beyond a 20K bandwidth ? Is it the same reconstruction of the actual musical event? at DAC level? at HQP level?

 

In other words : OK it's going to be rolled off starting at 20K but do HQP and/or the DAC do the same reconstruction and output signal of the same quality wether they are fed with relevant real musical info beyond 22K or not ?

Link to comment
What are the changes in 3.16?

 

The link to the "Version History" page is the "Main" button in the left column of the web page.

 

The download link is near the bottom of the page linked by the "Consumer" button.

HQPlayer (on 3.8 GHz 8-core i7 iMac 2020) > NAA (on 2012 Mac Mini i7) > RME ADI-2 v2 > Benchmark AHB-2 > Thiel 3.7

Link to comment

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...