Do I take everything out now? Or is there something that I dont know

My convultion is from AutoEQ

And my system is HD660S + Pro-ject Pre Box S2 Digital

One more info is that sdm is -3db and headroom 0db just to see if its clipping. So the 2nd pic was little bit wrong.

Hi @Koffi, it depends on what you’re looking for. It looks as though the convolution filter and the DSD upsample combined is too much (looks single core bound). I use similar headphone filters from the same site and they generally don’t take much CPU. I don’t do DSD but understand that it’s quite CPU intensive.

Personally I’d drop the DSD and stick to the headphone filter. Alternatively, given your 1.9x rate all in, DSD alone should be fairly comfortable. If you prefer the sound of that without the EQ filter then try that? What about a less ambitious DSD rate, though again am not a DSD listener so just speculating.

What do you mean it is though? What I’ve understand everything above 1.0 is ok. And this is my laptop so I thought 1.8-2 is ok?


Its not even using 50%

You are right, 1.9x should be OK, it’s just getting quite close to problems. I was assuming you had a problem. As long as the rate is consistently higher than 1x you don’t have a problem.

I dont have any problems Im just asking is there something more to try :smiley:

Not if you want that rate to stay above 1x :wink:

I think there is still lot of “headroom” with 1.9. :face_with_raised_eyebrow:

edit. Because its running tv and everything same time without problems.

You are right in theory, it’s twice the speed you need. That’s assuming that the processing rate is constant. I’d keep the headroom and enjoy listening with no pauses or skips.

I assumed you had an issue and am happy to help. I’m not the right person to come up with further tweaks but I’m guessing it already sounds pretty nice? Enjoy :slight_smile:

Regarding

you have plenty of spare CPU with the other cores but I’m pretty sure that Roon DSP is one core per endpoint so won’t be able to use all of that.

I understand what you mean, but my buffer is 1.18ms already so I dont have any “pauses or skips”. It sounds good and I already paid 1y license for roon after many trials :sweat_smile:

And its not one core per endpoint.ddddd

edit. I just think 1core of my 9750h can handle it pretty well so It doesnt even need more cores. Because I cant get even fans to run more with Roon

But this make me think, what the hell you do with nucleus? If you have ok pc and you know what youre doing you dont get anything from it or can somebody open this for me? :smile: I have server at home but I think I dont get anything using it for this


Can somebody explain this? I have HIFI chosen at services, No MQA support and disabled mqa decoder.

Why this is still playing 24bit mqa?

That’s the source file, i.e. 44kHz 16bit. No MQA processing is performed. The bit depth conversion is needed for accuracy prior to converting PCM to DSD.


Why then no MQA is like that?

The source file isn’t MQA in this example. Let’s Welsh through the first signal path.

  • Source file is MQA 44.1kHz 16bit, but note that MQA stores the additional information in a 16 or 24bit wrapper, but for all intents in this example they contain no data [since you have disabled Roon’s MQA core decoder].
  • MQA file authenticated (note that there is no deciding following this step, i.e. unfolding to 88.2kHz.)
  • Bit depth conversion to 64bit floating point. As mentioned this is for accuracy.
  • Sample rate conversion since DSD is 1-bit with high sampling rates.
  • DSD uses sigma delta modulation; the output is a bit stream.
  • Bit perfect output from your Roon Core
  • connected device is the Pro-Ject Pre Box S2.

It may be worthwhile checking that your DAC truly supports DSD and isn’t converting this back to PCM.

I mean there should be FLAC and MQA files in Tidal. MQA should not be there anymore.

And for DSD I think I believe Miska: