HQPlayer Desktop thread

Yes, device selection in HQPlayer stores device id which to use. For NAA endpoints this is always combination of two things - NAA name and device id of the DAC behind it. This is also reason why each NAA must have a unique name, otherwise HQPlayer will get confused.

It is normal to have multiple DACs behind a single NAA, or have multiple NAAs with one or more DACs behind each. I do it too.

It can now with new firmware update

1 Like

Wow - did they really do it?! Or its just a marketing nonsense that no one in iFi understands but it sounds so freaking fancy though !?!?

Anecdote #1 my ticked with iFi about Go Bar DSDx48 rates is still open - precisely two months with no single indication of an answer. Front line has apologised a month ago for “delayed” response. I personally find that pathetic, but, hey, this is the level of service you must expect with iFi. Pure nonsense… I have returned their devices weeks ago.

Now for the Gryphon I’m very sure its a different animal to a nonsense-dongle and if this can really perform fully fledged it may actually be something!!!

Anecdote #2 I have been very excited to prepare my “summer break” HQP station, and am very happy as I really got impressive desktop setup, but… At the end I was so freaking tired before the break that I did not take any sound source/music device with me. However now i have fully immersive 3D audio of Cote d"Azure hot breeze 24/7 instead and I find this exactly what I need right now blended with good chilled wines or better - Pastis :slight_smile: Because back at home there are plans of changing cooling system on my HQP 129k server and making a final assault at 7ECv2@1024 ! :))) May be those extra 5-6C i count on will make a difference??? Igorski - it is really a time for a break! hahaha!!!


apologies if this has already been answered elsewhere, but for an M1 Mac Mini, is 8gb of ram sufficient for HQPlayer? Or does it need 16gb?


For simpler use cases it is sufficient. For complex ones maybe not.

But even without HQPlayer, 8GB models suffer from early flash (SSD) wear out due to increased swapping.

Good afternoon everyone. as I said before, I like the work of hq player, now I have a Gustard x26 pro DAC, as a source a laptop with an intel i5 processor, in the player settings is adaptive output rate, sin-m filters, dsd 48 × 256 conversion , and asdm7 modulator

Jussi, could you please explain roughly what you see as a simple use case and where it is going to be a complex use case?


P.S. I am obviously also a MacMiniM1/8GB user. :smiley:

Due to the fact that I want to use more advanced regulators, I plan to change my computer, but I don’t know which processor and video card to buy specifically, because of this I experimented with my friend’s hq player. He has an Intel Core i7-9750H and GeForce RTX 2060 SUPER 8G. So that’s interesting without connecting cuda, his laptop could only reproduce the settings as much as possible on my i5, the music stuttered on more complex modulators, although in the task manager window the CPU load did not exceed 35 percent, but when I connected the video card, it had GeForce RTX 2060 SUPER 8G , then the maximum that I could get was asdm7 , sin-m and 48x512 . In this regard, I have a question, what is more important when assembling a new PC for hq player? processor or video card?

what do I do next - I take my DAC and go to a friend who has a Processor - Intel Core i7-11700K 3.6(5.0)GHz 16MB s1200 Box (BX8070811700K)

Video card - Asus TUF GeForce RTX 3060 Gaming OC 12288MB (TUF-RTX3060-O12G-V2-GAMING). So, asdm5ec 5ecv2, 7ec, 7ecv2 modulators work only when cuda is enabled, sin-m filters, 48×256 conversion, adaptive output rate .(48×512 stutters) , CPU usage 33 percent. That is, again it turns out that there is no way without a cool video card, or am I wrong? I just wanted to give all the work to the processor, so as not to buy an expensive video card …Jussi please advise processor and video card for asdm5ec , 5ecv2 , 7ec , 7ec2 modulators and 48x256, 48x512 conversions.

Hi Igor

I am running DSD512x48 with ASDM7ECv2 (auto family rate enable)

1x = gauss-long
Nx = gauss-long-hires-lp


My i9-9900K does it fine. And my i9-11900K of course.

But I am running HQPlayer Embedded running the multicore optimised build Jussi has created, running on Ubuntu Server.

I don’t have seperate GPU.

You already have full Desktop license? Consider Embedded?

1 Like

Happy for you ! i9 is cool! , I will assemble a PC from Windows, because I will have not only hq player server there, but also roon core and music and movie storage, but I feel it will not be soon …that is, I’m targeting the desktop version

1 Like

So what kind of machine are you planning to build?

CPU model? What cooling? Motherboard model? Power supply? Case?

You may not even need the GPU. Read through the recent posts in the HQPlayer Embedded thread. Users have had really good results with i9-12900k builds but they are on embedded and outputting DSD1024.

I bet you will have more than very good results with a i9-12900k, even without a GPU, using HQP desktop.

13th gen chips will be out soon too. Maybe wait for one of those, z790 board, DDR5 ram… (i’m drooling now… :drooling_face:)

you’ll spend more on the processor and ram but you’ll save by not needing the GPU.

1 Like

First I’m thinking of buying sotm sms 200 ultra to use as NAA.
And then I’ll probably take something from the top i9 line and a motherboard, and then we’ll see …the PC case will be simple, cooling to the maximum, since it will not be in the listening room

Simple case: stereo upsampling with poly-sinc filters
Complex case: Matrix processing with multi-channel output with convolution


How about stereo upsampling with poly-sinc filters and convolution? :wink:
I think this is going to be my most likley use case.

If you already have a machine with 8gb RAM, you’ll be fine with HQPlayer with poly-sinc filters.

When you next buy/build a new machine, then get 16db RAM for the reason Jussi mentioned before


GPU can offload filters and convolution, but not modulators. It can free up CPU cores for processing modulators in multichannel case. Or reduce load on CPU cores to allow higher turbo boost clocks.

GPU doesn’t directly affect these. Only through CPU turbo boost, which in turn depends on load of multiple cores. It works by boosting clocks of few highly loaded cores if there are enough CPU cores on lower loads.

So your filter choice and and offload status affects which modulators you can run.

If you have a fast enough GPU, it may also allow you to use heavier filters. If your GPU is too slow, CPU ends up waiting for it to complete it’s tasks.


question to Jussi and other members, who knows. Do I need sotm sms 200 ultra as NAA? Or can I use some kind of intel nuc?


NUC can work.

But Raspberry Pi 4 works well. So does this. Looks like in stock and shipping.
Both are powered by 5Vdc

1 Like