Jump to content
IGNORED

HQPlayer Linux Desktop and HQplayer embedded


ted_b

Recommended Posts

1 hour ago, lpost said:

 

Perhaps it's the dynamics. The punch and pop. The waves of sound are softer, less impactful, for me with DSD256. I think of these as transients and perhaps I simply like the less accurate rendition.

 

My PC is not been able to run beyond 256 and would love to experience higher but I also don't want a ~300-500w monster sucking more energy than my 265w 120v system currently does.

 

I spent a few hours piecing together a 6 and 8 core spec for HQPeOS x64 but haven't moved ahead with it just yet. I've not exhausted my exploration of PCM.

 

I have the May L3 dac fed USB with UP Gateway NAA. My previous DirectStream dac only output DSD and I was limited to 192kPCM by I2S board.

 

I've not spent much time with the numerous combinations available for DSD. Any suggestions where to start?

 

1998786596_ScreenShot2021-06-13at10_50_58AM.thumb.png.ad203a1153c700fda5300db0ebfbeb08.png

 

Use same filters for DSD as you use for PCM. And use ASDM7EC modulator for Holo DACs.

 

But also remember to turn up volume 6 dB higher when you play DSD, because the DAC has 6 dB lower output volume for DSD sources. Or alternatively turn HQPlayer volume 6 dB lower when playing to PCM output to compensate for the difference.

 

My both Spring and Spring 2 sound more flat with PCM inputs. Like bass drum is "flap flap" with PCM instead of deeper "thump thump" with DSD. With 32x input rates the flatness is much less than at lower rates, but still the flavor is there. And cymbals are cleaner with DSD than with PCM input, PCM inputs sound more congested. To me, DSD can keep things separated and less congested even when there's a lot going on simultaneously.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
15 minutes ago, Miska said:

 

Use same filters for DSD as you use for PCM. And use ASDM7EC modulator for Holo DACs.

 

But also remember to turn up volume 6 dB higher when you play DSD, because the DAC has 6 dB lower output volume for DSD sources. Or alternatively turn HQPlayer volume 6 dB lower when playing to PCM output to compensate for the difference.

 

DSD128 has a terrible fuzz in the central image and DSD256 will run for about 12 seconds before a dropout. 256 is nice but too nice. The punch is wimpier. The image more diffuse. I could adjust the speakers for it but only if it will play on. Both with ASDM7EC. So I stick with PCM for now and look forward to exploring DSD if/when I build a newer machine.

 

CPU hardware has not made the tremendous strides it used to. That darn <10nm barrier. GPUs are as good as not available. And need more power!

 

Too bad Holo didn't account for the 6dB gain delta.

Link to comment
18 minutes ago, lpost said:

I'm just a touch less at i7-6700 or 6770 I forget which. It's a retired business computer from my office. 3 years old and they go. Perhaps I should set my sites on a decent 8th or 9th gen CPU and not 11th.

Just get a MAC Mini M1 and use it just for HQP. I can get ASDM7EC/DSD 256/48/poly sinc long gauss with no problem. And, it is an economic al solution. Get the 16 GB version.

Link to comment

I would need to purchase desktop license as I don't believe HQPe will run on Mac. I really don't want to install and manage Ubuntu on my audio PC. I get plenty of it with my work. I am grateful Jussi puts together the bootable images. Bam, easy. Done. Swap anytime back to previous build and move forward without risk to music flow.

 

I need a CPU, MB, RAM, case and I have a power supply I could reuse. I've looked at i5-11600k and i7-11700k and a mid-level board with decent VRM. Ram is not a huge deal as I don't believe overclocking it a ton will really improve HQPe. There are some cool cases and I have the space to fit an ATX in my AV closet. Fans are zero issue.

 

Frankly, the May sounds so very good NOS and HQPe NOS with LNS15 as I am using volume I wonder the value of a new PC when what I've got gets me so very close. 1.5M PCM does bring a touch more focus, width and depth et al but I don't have to have it to enjoy the music.

 

When will a CPU and/or GPU combo be able to do DSD512 with EC modulator, let alone 1024, that doesn't make a decent heater. I'm electricity neutral or even in a surplus with my 5kw solar. It's not about the $ it's more about saving resources. 

Link to comment
1 hour ago, lpost said:

I would need to purchase desktop license as I don't believe HQPe will run on Mac. I really don't want to install and manage Ubuntu on my audio PC. I get plenty of it with my work. I am grateful Jussi puts together the bootable images. Bam, easy. Done. Swap anytime back to previous build and move forward without risk to music flow.

 

I need a CPU, MB, RAM, case and I have a power supply I could reuse. I've looked at i5-11600k and i7-11700k and a mid-level board with decent VRM. Ram is not a huge deal as I don't believe overclocking it a ton will really improve HQPe. There are some cool cases and I have the space to fit an ATX in my AV closet. Fans are zero issue.

 

Frankly, the May sounds so very good NOS and HQPe NOS with LNS15 as I am using volume I wonder the value of a new PC when what I've got gets me so very close. 1.5M PCM does bring a touch more focus, width and depth et al but I don't have to have it to enjoy the music.

 

When will a CPU and/or GPU combo be able to do DSD512 with EC modulator, let alone 1024, that doesn't make a decent heater. I'm electricity neutral or even in a surplus with my 5kw solar. It's not about the $ it's more about saving resources. 

The upcoming M1X/M2 chip might be the first to be able to do ASDM7EC/DSD512. We will see this winter.

Link to comment
1 hour ago, bobflood said:

The upcoming M1X/M2 chip might be the first to be able to do ASDM7EC/DSD512. We will see this winter.

 

I think it would really be something if there was a way to access the "neural engine" of the chip and use that to assist with the upsampling/filtering.

No electron left behind.

Link to comment
7 minutes ago, AudioDoctor said:

I think it would really be something if there was a way to access the "neural engine" of the chip and use that to assist with the upsampling/filtering.

 

Neural engines typically use 8-bit integer and 16-bit floating point math, which is something unsuitable for audio... While for audio we need 64-bit integer and floating point. And for some cases also 80-bit and 128-bit floating point.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
6 minutes ago, Miska said:

 

Neural engines typically use 8-bit integer and 16-bit floating point math, which is something unsuitable for audio... While for audio we need 64-bit integer and floating point. And for some cases also 80-bit and 128-bit floating point.

 

 

Thanks for clarifying Miska, I had no idea,.

No electron left behind.

Link to comment

Have to admit all the oversampling options with HQPlayer are a bit daunting. I find I prefer PCM oversampling for both PCM and DSD with an Ares II DAC, running at its limit

of 705/768. Have followed a configuration  recommendation on Audiowise's site but do have a question

 

https://cdn.shopify.com/s/files/1/1552/8145/files/SRC-DX_HQP_Settings_3.pdf?v=1612442181

 

Am using Sinc-L for the 1x filter and Nx filter settings w/ LNS15 dither, adds a bit if delay for start of play + have to rate limit the CPU a bit higher for DSD but sounds really good.

Is this an Ok filter combination or is there something here that isn't good practice?

Regards,

Dave

 

Audio system

Link to comment
8 hours ago, davide256 said:

Am using Sinc-L for the 1x filter and Nx filter settings w/ LNS15 dither, adds a bit if delay for start of play + have to rate limit the CPU a bit higher for DSD but sounds really good.

Is this an Ok filter combination or is there something here that isn't good practice?

 

What do you mean by rate limit the CPU?

 

It is OK combination as long as apodizing filter is not needed for the source content (see the counter value). You get notably lower delay with poly-sinc-gauss for example, if you'd prefer lower delays.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

Link to comment
2 hours ago, Miska said:

 

What do you mean by rate limit the CPU?

 

It is OK combination as long as apodizing filter is not needed for the source content (see the counter value). You get notably lower delay with poly-sinc-gauss for example, if you'd prefer lower delays.

 

8i7 NUC BIOS allows you to set min and max frequency limit for CPU. I get cleaner sound running at or below 1200mhz, had to bump up

from 600 to 1200, max is 2700 with turbo (burst) mode disabled. Had to do same when I tried DSD256 up-sampling

 

Thanks for the poly-sinc-gauss suggestion.

Regards,

Dave

 

Audio system

Link to comment
4 hours ago, lpost said:

How important is hyper threading to HQPe performance? 5% increase?

 

Depends on what else you have running on the system. It saves half of the multitasking context switching overhead of the OS and helps cache efficiency.

 

4 hours ago, lpost said:

How about AVX-512? I presume it's not yet used but will be in the future?

 

AVX-512 has been supported on Windows, macOS and generic Ubuntu Linux builds for couple of years.

 

Signalyst - Developer of HQPlayer

Pulse & Fidelity - Software Defined Amplifiers

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...