# Maximum Computational Settings in HQ Player

**URL:** <https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318>\
**Category:** HQ Player\
**Created:** [July 6, 2020, 3:08am UTC](https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318 "2020-07-06T03:08:46Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![andybob](https://community.roonlabs.com/user_avatar/community.roonlabs.com/andybob/32/109007_2.png) [@andybob](https://community.roonlabs.com/u/andybob)\
**Post date:** [July 6, 2020, 3:08am UTC](https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318/1 "2020-07-06T03:08:46Z")

</div>

Hi @jussi_laako,

I know you’ve mentioned from time to time the single core performance required for the EC modulators at higher sampling rates (9 GHz for DSD 512 iirc).

In this thread I wanted to talk about the most intensive current settings in HQPlayer and ask your opinion about what hardware spec would be required to run it and when such computing power might become available.

The maximum settings I am thinking of are:  
192 kHz/24 bit source file  
Poly-sinc-xtr-lp filter  
ASDM7EC modulator  
48k x 2048 DSD bit rate

Obviously no current domestic computer could run these settings, but is there _any_ current computer that might do so ? For example, I understand [Fugaku](https://www.top500.org/lists/top500/2020/06/) is currently the “fastest computer in the world” at 415.5 petaflops. Could it run HQPlayer with such settings ?

Second question. Are there any aspects of audio DSP which might benefit from quantum computing ? Would performing operations on a superposition of taps save time ?

---

<div class="post-metadata">

**Author:** ![jussi\_laako](https://community.roonlabs.com/letter_avatar/jussi_laako/32/5_5575768a8748004e209b776fc1b2916d.png) [@jussi\_laako](https://community.roonlabs.com/u/jussi_laako)\
**Post date:** [July 6, 2020, 3:20pm UTC](https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318/2 "2020-07-06T15:20:37Z")

</div>

> [@andybob](#):
>
> I know you’ve mentioned from time to time the single core performance required for the EC modulators at higher sampling rates (9 GHz for DSD 512 iirc).

There’s a benchmarking feature you can actually use to test this. If you select Null Output backend in HQPlayer settings, you can run such tests to estimate how much faster computer you need. This backend just throws away the processed data without doing any output. This allows processing to run at full speed. As usual, at the end HQPlayer reports elapsed playback time in statusbar. Comparing this playback time to the content length you get idea of the processing speed vs playback length,

You can try with both Multicore DSP set to auto (grayed) and set to fully enabled (checked). At the moment, only some AMD CPU models seem to perform better with fully enabled (maximum parallelization).

> [@andybob](#):
>
> The maximum settings I am thinking of are:  
> 192 kHz/24 bit source file  
> Poly-sinc-xtr-lp filter  
> ASDM7EC modulator  
> 48k x 2048 DSD bit rate

I would say 1024x DSD would be enough so far, since that is IIRC the maximum DACs today support.

sinc-L would be also fairly taxing one, but GPUs greatly help that one. Although for DSD1024 you may be running out of GPU RAM on all models.

> [@andybob](#):
>
> Obviously no current domestic computer could run these settings, but is there _any_ current computer that might do so ? For example, I understand [Fugaku](https://www.top500.org/lists/top500/2020/06/) is currently the “fastest computer in the world” at 415.5 petaflops. Could it run HQPlayer with such settings ?

Those are all massively parallel computers, they don’t have particularly fast individual cores. Just many of them. But not all algorithms can benefit from such massive parallelism. So for above case it would likely do worse than i9-10900K which is current king of per-core speeds (single thread performance).

Such machines would be however useful for quickly converting your entire music library offline. It could possibly convert several terabytes worth of music library in minutes.

> [@andybob](#):
>
> Second question. Are there any aspects of audio DSP which might benefit from quantum computing ?

I don’t know, I have to say I have not spared much thoughts on quantum computing so far. It is quite peculiar area of computing on it’s own. Very good for cracking crypto-algorithms or such. Not sure about audio.

---

<div class="post-metadata">

**Author:** ![Sebastian\_Tomita](https://community.roonlabs.com/letter_avatar/sebastian_tomita/32/5_5575768a8748004e209b776fc1b2916d.png) [@Sebastian\_Tomita](https://community.roonlabs.com/u/Sebastian_Tomita)\
**Post date:** [July 7, 2020, 11:58am UTC](https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318/3 "2020-07-07T11:58:29Z")

</div>

Do you think that the announced apple silicon would be a good solution for HQPlayer?

---

<div class="post-metadata">

**Author:** ![jussi\_laako](https://community.roonlabs.com/letter_avatar/jussi_laako/32/5_5575768a8748004e209b776fc1b2916d.png) [@jussi\_laako](https://community.roonlabs.com/u/jussi_laako)\
**Post date:** [July 7, 2020, 12:04pm UTC](https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318/4 "2020-07-07T12:04:42Z")

</div>

Sorry, I haven’t been following such closely. But if it is ARM based, it is likely not as good as Intel based ones.

---

<div class="post-metadata">

**Author:** ![Sebastian\_Tomita](https://community.roonlabs.com/letter_avatar/sebastian_tomita/32/5_5575768a8748004e209b776fc1b2916d.png) [@Sebastian\_Tomita](https://community.roonlabs.com/u/Sebastian_Tomita)\
**Post date:** [July 7, 2020, 12:28pm UTC](https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318/5 "2020-07-07T12:28:47Z")

</div>

> [@jussi\_laako](#):
>
> i9-10900K

It seems that the difference in single core performance is already tiny. And this is a mobile CPU.

> **[Intel Core i9-10900K vs Apple A13 Bionic - GadgetVersus](https://gadgetversus.com/processor/intel-core-i9-10900k-vs-apple-a13-bionic/)**
>
> Comparison between Intel Core i9-10900K and Apple A13 Bionic with the specifications of the processors, the number of cores, threads, cache memory, also the performance in benchmark platforms such as Geekbench, Passmark, Cinebench or AnTuTu.

---

<div class="post-metadata">

**Author:** ![jussi\_laako](https://community.roonlabs.com/letter_avatar/jussi_laako/32/5_5575768a8748004e209b776fc1b2916d.png) [@jussi\_laako](https://community.roonlabs.com/u/jussi_laako)\
**Post date:** [July 7, 2020, 5:02pm UTC](https://community.roonlabs.com/t/maximum-computational-settings-in-hq-player/113318/6 "2020-07-07T17:02:54Z")

</div>

> [@Sebastian\_Tomita](#):
>
> It seems that the difference in single core performance is already tiny. And this is a mobile CPU.

It is not tiny, what I’ve compared to other 64-bit ARM cores (Cortex-A53), they are notably slower clock-for-clock than Intel’s Atom cores. Not to even mention something like Core or Xeon models that have AVX2 and AVX-512 and similar. Cache sizes and memory bandwidths tend to be also notably different.

So far fastest quad-core 64-bit ARM I have can just about do DSD64 upsampling. While fastest quad-core Atom I have can do DSD256 upsampling. Both with similar TDP.

If some test shows a small single core performance difference, then the workload is not anything similar to HQPlayer’s DSP pipelines.

We’ll see, but I’m not holding my breath on performance expectations.
