I’ve bounced back and forth between Linux flavors a lot and lately I’ve been using Linux Mint with Cinnamon desktop. The setup I’m running now I’ve found to be fairly simple in the maintenance dept., and your Linux background and experience appears to be much more advanced than mine.
Current rig is an old rig I previously used for a hackintosh, consisting of a Z490 Vision D, core i9 10900k, 64gig 3200 ram, nvme drive, and an old gtx 1080 that someone gave me.
I’ve been running hqplayer desktop and RoonServer on this machine. It’s set to auto log in at boot and launch hqplayer.
I manage it with xrdp and cockpit. Most of the time I use an iPad with Screens to access Remote Desktop, a web browser for cockpit and terminal for other misc stuff.
Linux Mint now has better support and easier installs of nvidia drivers too via a Driver Manager settings panel.
It does mostly everything DSD256 (some utilizing -2 variants)
A lot more capable than the M1 Max I used previously as the intel setup with cuda offload will also get me into a lot of DSD512 and allow me use of 7EC-Super which the M1 refused to run without dropouts. I have a rtx3080 ftw but it’s a bit overkill and a lot more hungry in the power dept. I find that DSD 256 with adaptive rate and 48k checked, is my goto with the May KTE.
My setup is quite simple now: jussi recommended up board with NAA OS → MayKTE → SereneKTE → McIntosh MC462 → 803D
I will most likely stick with Mint for a while.
I use jussi’s hqplayer client on iPad to change settings while listening.
It’s very powerful, congratulations. One question: how do you install the optimizations Jussi was talking about? Is there a repository with deb packages for Jussi’s optimized kernels?
Question is what kind of impact those have. My Ubuntu build is built, tested and optimized for Ubuntu. And Fedora build is built, tested and optimized for Fedora. I’m personally completely happy with the performance. In particular I use the distro versions officially supported by Nvidia for CUDA, since it is important factor for me. HQPlayer OS doesn’t support CUDA, since it is not possible there for multiple reasons, it is pure CPU-only.
No repository, you can just download and install the kernel packages as you feel fit.
About processors: we find more and more computer, often laptops under Win 11 with a Qualcomm Snapdragon X.
X86 and X64 softwares are emulated ?? And for the installation of HQPlayer or NAA, a solution?
I still have not seen any other such devices than few Microsoft Surface tablets.
Windows has been so long tied to x64 architecture, that arm64 is still a mess there. There are few Windows related parts that would require some extra work from my side. I would need to have some decent development hardware to begin with such, and that is certainly not something like Microsoft Surface.
But at the moment, if you’d like to use such arm64 hardware with native support, I would recommend Linux, or macOS.
Yes, for the moment, it’s very reduced.
I am not sure that this is developing beyond laptops who are just more powerful than smartphones.
No need to HQPlayer but perhaps a specific desktop version or a specific NAA package to make an endpoint while remaining under Windows.
HQPlayer Client is pretty easily doable, I would just need a suitable development/test device. HQPlayer server side is more complicated and probably not very huge percentage of HQPlayer users.
I don’t really understand why NAA would be useful there. In particular because I have never seend any DAC manufacturer provide arm64 ASIO drivers. For NAA, it is much better option to just boot Linux there and install the already available arm64 NAA package there.
I don’t know how to buy any Dell stuff in Finland, I don’t think it is generally available here.
Hi @jussi_laako My iFi iDSD Pro (not Signature) has arrived today and I’ve spent the whole afternoon testing it with HQPlayer Desktop 5. The sound is very good, but I get glitches from time to time. I’m using DSD Direct 256 / Shaper: ASDM7EC-fast / Filter: poly-sinc-hb-I-2s / Integrator: IIR / Conversion: XFi. What changes would you recommend to eliminate these glitches? Thanks for your help.
Is Limited counter staying 0 at all times? What kind of source format gives you the glitches?
Of course you could also be just periodically running out of processing power due to some other activity in the computer, if you are close to it’s performance limits.
Thank you for your reply. The Limited counter remains at 0. There are many apods. The source is FLAC 44.1. Processor activity never reaches 10% with this configuration. The temperature ranges between 45º and 90º.
Thank you, Jussi, for your reply. I had the fixed volume box checked. I changed it to -3 dB (even to -6 dB) and the glitches continued to appear. Regarding the apodizing filters, I have tried several these days and the glitches have not disappeared either. I even switched to PCM, with negligible processor load and temperature, and the glitches continued. In any case, please recommend a filter for DSD256 that I can use to perform more tests. Regarding the question of which core has the highest load in the graphs, sometimes it is 0 and 2, and then 1 and 3, but I don’t see them going above 60%. The CPU has 16 cores, and most of them remain almost inactive. It’s a NUC i7-1260P with 64GB of RAM, not very powerful. The overall load doesn’t exceed 11% with ASDM7EC-fast and poly-sinc-hb-2s. I’m including some screenshots:
Thanks, Henry. No, I hadn’t made that correction in the BIOS, but I looked up the instructions, did it, and the temperature seems to be the same, 50º-60º with peaks of over 80º. The server gets very little use. Practically only Roon, HQPlayer, and a couple of lightweight apps.
Depending on iFi model, you can maybe set “48k DSD” to checked, and hence also “Adaptive rate” to checked. Here assuming it is one of the DACs based on TI/BB chip. With the Cirrus Logic chip models YMMV.
From the ASIO driver control panel (in Windows System Tray), set buffer size to maximum possible and enable “Safe mode”.
If “48k DSD” is enabled, you can also set “Bit rate” to “48k x256”.
I have a question regarding the settings for HQPlayer Desktop.
Currently upsampling all my music (PCM/DSD) straight to DSD512 using the following settings:
With this configuration (DSD512), there is definitely more air, soundstage is clearly wider, quite a bit of that normal ‘digital’ glare / grittiness is gone, but noticing something interesting.
There is something new introduced up top…not sure if it’s considered oversampling noise, but something I can’t quite place my finger on. It’s quite subtle, but think it almost leads to listener fatigue over time. Will need to get some of my audio friends to have a listen sometime as this could just be psychoacoustic in nature.
Seems like some of the fast transients previously audible when playing PCM now feels slightly smoothed out. Seems like less PRAT perhaps?
Again, nothing conclusive here, just some general observations.
Could be settings I suppose. Therefore seeking guidance based on my settings as noted.
Machine specs
Processor Ryzen 5 5700F with 64 Gb of RAM
RTX 3090 Ti with 24 Gb of RAM
Not sure if this helps, but here’s the current Digital Architectural blueprint
I’ve heard on this thread leveraging a DAC capable of handling pure DSD would be beneficial, but again seeking your valuable input, as I think the Laiv Harmony converts everything pack to PCM internally which may also influence the sound in a negative way?
When I hit the limits of my NUC in Akasa Pluto with TDP strapped to 28, I get two seconds or so of playback followed by a second pause over and over. That is with an i71360P (Asus NUC PRO). It happens before indicators like temperature and CPU usage justify throttling. If you cannot roll back to settings where the NUC behaves itself then there is something else wrong. But one thing I haven’t done is update my bios. Later gen 13 and 14 bios settings prevent a bug that kills some processors and the motherboard they use and I am pretty sure the updates limit the voltage demands of the CPU.
