[bogi]

So what will be better a longer filter like sinc LL and pcm to pcm upsampling or a lighter filter with modulator to convert pcm to dsd?

Maybe you told it in the past but I forgot what's your DAC.
There is no such equation (at least theoretically) that highest quality result bring longest and hardest to process filters. We discussed it some weeks ago in this thread. But of course, some users may like most just those.
Others may have different meaning, but on your place I would go the PCM to DSD route with HQPlayer (if you have suitable DAC available) and I would start with middle length filters which are good compromise from many aspects. You may look at filters I suggested in my previous post.

 

[rkt31]

Maybe you told it in the past but I forgot what's your DAC.
There is no such equation (at least theoretically) that highest quality result bring longest and hardest to process filters. We discussed it some weeks ago in this thread. But of course, some users may like most just those.
Others may have different meaning, but on your place I would go the PCM to DSD route with HQPlayer (if you have suitable DAC available) and I would start with middle length filters which are good compromise form many aspects, like those I suggested in my previous post.

My dac does not deal dsd natively.

 

[bogi]

My dac does not deal dsd natively.

The term 'native DSD' was in the past used as opposite to DoP for USB transfer mode. The term 'direct DSD' was used in the meaning of direct DSD path existence (direct DSD may work with DoP too if DAC supports it). But people started to use the term 'native DSD' for direct DSD path too and that makes unclear what is the speak about. What for DAC you are using?

 

[rkt31]

The term 'native DSD' was in the past used as opposite to DoP for USB transfer mode. The term 'direct DSD' was used in the meaning of direct DSD path existence (direct DSD may work with DoP too if DAC supports it). But people started to use the term 'native DSD' for direct DSD path too and that makes unclear what is the speak about. What for DAC you are using?

I know dsd transmission as dop and without dop ( called native support) is also used by many. But my chord dac concerts dsd into pcm.

 

[bogi]

Chord DACs at first resample everything (any input PCM/DSD content) to 705.6k PCM and then apply their own processing. With Chord DACs it is recommended to set PCM output and 705.6k output sample rate. That's the best what you can do. Chord DACs are not optimal for use with HQPlayer, but you may still perceive an improvement that way. Just try it. You have to find a HQPlayer filter which you like most. Some suggestions also for dither (noise shaper) setting you can find in the following links:

https://audiophilestyle.com/forums/topic/19715-hq-player/?do=findComment&comment=1233487
https://audiophilestyle.com/forums/topic/19715-hq-player/?do=findComment&comment=1232658
https://audiophilestyle.com/forums/topic/19715-hq-player/?do=findComment&comment=1108209
https://audiophilestyle.com/forums/topic/19715-hq-player/?do=findComment&comment=1065097
https://audiophilestyle.com/forums/topic/19715-hq-player/?do=findComment&comment=1150163
https://audiophilestyle.com/forums/topic/19715-hq-player/?do=findComment&comment=1164766

 

[genefruit]

I know dsd transmission as dop and without dop ( called native support) is also used by many. But my chord dac concerts dsd into pcm.

If you're using a Chord DAC, stay with PCM

 

[Superunknown2]

If someone was going to build a PC to run HQ Player out of, where they could run anything HQ Player has without trouble, what kind of CPU, GPU, ram, cooling etc would they need?

 

[genefruit]

If someone was going to build a PC to run HQ Player out of, where they could run anything HQ Player has without trouble, what kind of CPU, GPU, ram, cooling etc would they need?

I'm not sure there is one. I crosswalk what you've requested into 1x/nx ASDM7ECv2/3 at DSD1024x48 using Sinc-Mx or Sinc-L, with rate family unchecked. I'm very close with my 14900k/4080 using PGGB to perform offline upsampling to PCM 7XX then using SDM upsampling to the above. Even with those measures, I can't do the above with Sinc-Mx. @EMINENT any thoughts?

 

[EMINENT]

No matter. Just to easier find the right modulator it is suitable at first to set some high quality but not too demanding oversampling filter, like gauss-long or gauss. Or for hires content some of gauss modulators which contain 'hires' in their name. Or some of poly-sinc-*-2s filters, best to choose middle length (not short, not long) for the first try. Simply choose one and search for a modulator you like most.



sinc-L or sinc-long are quite hard job for a computer without nVidia GPU and long filters have also some disadvantages (transient smear with source content containing fast transients), so I'm not sure it has sense to invest $$$$ to be able to run sinc-L, when there are so many other possibilities. I like more middle length filters like poly-sinc-mp, poly-sinc-hb-m, poly-sinc-gauss-halfband, poly-sinc-gauss and with some content (like prog rock of 70's) poly-sinc-short-mp. Just some examples. They are easier on resources than sinc-L and transient smear with 'fast sounds' is not present with them.

If somebody is able to distinguish between sinc-LI vs sinc-L and sinc-L considers as the best one worth of investment, then of course, an investment to hardware makes sense. One can try to compare sinc-Lm vs Sinc-LI vs sinc-L at DSD256 rate and based on that to decide, if it has sense to upgrade hardware for sinc-L to be reachable at 512 (or 1024).

@EMINENT has setup able of running the hardest load filters, he can help on this topic more.

So what will be better a longer filter like sinc LL and pcm to pcm upsampling or a lighter filter with modulator to convert pcm to dsd?

Hullo.

I am no better than anyone here to make recommendations.
The answer to which those seek is up to what your setup sounds like to you. IMO, regardless of what is technically or measurably better, I will always follow my own ears and body sensations. I know this does not apply to everyone, so sonically I would not be caught up with just using only sinc MX or sinc Li or LL if you are cpu limited. If you need more warmth, try an EC modulator at a lower rate but try to stay 256 rate and above. There are many to enjoy without EC as well as I always found greater enjoyment at higher rates without EC vs stepping down with. YMMV.
Many do sound similar but a few will hopefully standout depending on your dac, you may like PCM instead or as some pointed out even NOS.

I'm not sure there is one. I crosswalk what you've requested into 1x/nx ASDM7ECv2/3 at DSD1024x48 using Sinc-Mx or Sinc-L, with rate family unchecked. I'm very close with my 14900k/4080 using PGGB to perform offline upsampling to PCM 7XX then using SDM upsampling to the above. Even with those measures, I can't do the above with Sinc-Mx. @EMINENT any thoughts?

It appears going from higher rate PCM>DSD 1024 is much easier than from 44.1/48. I am able to sustain PGGB 1.411 MHZ files upsampled to DSD1024 with sinc L but not sinc MX either. Another user on other forum with bigger case and cooling is able to do this I think with a $5k A6000 with 48GB ram if you wish to explore that option.

For me, I have never aspired for sinc L at any rate. Even 1.411 PCM files upsampled to 1024. If sinc L is your goal from 44.1/48 rates to 1024, it may be possible with the A6000/14900k combo with higher overclock than 6.1. It was not possible with 13900k/4090 in my same case and cooling though.

 

[b0bb]

Hi
Im looking to do dsd1024 on a Holo may. Would you recommend the 7950x3d or would you change to Intel 14909k. I have a 4090 GPU.

HQPlayer 5 uses the AVX-512 vector floating point instructions, Intel removed this from retail processors like the 14900 series.
AMD added AVX-512 support in Zen4 and is one of the reasons I chose AMD. I bought the 7950X3D
Uplift in performance is about 20%
AVX-512 on Intel is reserved for the Xeon enterprise class range, you end up with too many lower clocked cores and high cost.

If you are using GPU off load, the load profile is quite unique.
Only 2 cores are heavily loaded, Zen4 does well here, holding steady at 5.4GHz, HQP5 fixed the core pinning issues.
I use DSD512, you should ask around for DSD1024 performance numbers

Below example is a trace of HQP5 SincLong 192k PCM to DSD512 conversion. Linux 6.2 low latency kernel
On equivalent clocked Zen3 without AVX-512 the load is about 85%

 

[Louisiana]

New version 5.31 is out - now you can change filters/dithers on the fly.

Edit:
To change filters/... on the fly, you need the HQPlayer Client

 

[alxw0w]

I have a question regarding license for HQPlayer.
Is the license transferable through different computers?
Currently I would like to use HQPlayer on my main workstation but in the near future I'm planing to use Mac mini or other pc workstation.
No problem with transferring license through different instances?

 

[Louisiana]

No problem with transferring license through different instances?

No problem, but you can only use one instance of HQPlayer at the same time.
If you run it on Win PC, can’t use it on Mac, and vice versa.

I use one license for both, my Mac Mini M1, and if I need some extra horsepower I use it on my Windows PC with RTX4090 GPU

 

[Superunknown2]

I have been using HQ Player through Roon on my Macbook Pro. The Roon is set so I can choose in the bottom right of the screen either my Holo Spring 3 DAC (for NOS) or HQ Player (which would oversample and then send the signal to the Spring 3 DAC). Usually when I use the Spring 3 option for NOS in Roon, and then change Roon to HQ Player, I have to close out of Roon and HQ Player, and restart them before I can use HQ Player correctly. If I don't, then the sound will come out of my Macbook speakers. Is there some setting that I can change so I don't have to restart HQ Player and Roon when I am doing this?

 

[b0bb]

I have been using HQ Player through Roon on my Macbook Pro. The Roon is set so I can choose in the bottom right of the screen either my Holo Spring 3 DAC (for NOS) or HQ Player (which would oversample and then send the signal to the Spring 3 DAC). Usually when I use the Spring 3 option for NOS in Roon, and then change Roon to HQ Player, I have to close out of Roon and HQ Player, and restart them before I can use HQ Player correctly. If I don't, then the sound will come out of my Macbook speakers. Is there some setting that I can change so I don't have to restart HQ Player and Roon when I am doing this?

Latest 5.3.2 version of the desktop app is better behaved when it comes to sharing the same audio output device with Roon.
On my own setup getting Roon and HQplayer to share the same output device has been an uneasy existence.
Using separate output devices on the DAC for example USB for HQplayer, SPDIF or I2S for Roon will address the issue if it works for your use case.