I stream DSD1024 and have been getting wonderful results with AHM7EC8B using the T+A DAC200 but can I improve with AHMxEC4B? I’m using embedded headless Linux.
Yes, it is better choice in particular for the DAC200. I also recommend to keep the DAC200’s 60k analog output filter enabled and matching DAC correction enabled for best performance.
Ok thanks. Looks like this might be a game changer for me. I assume the NAA needs to be upgraded at the same time
NAA is not so much related to this, as long as it is relatively up to date to enable DAC corrections (something like >= 5.1.5).
Ok thanks again.
The upgrade has been successful and new licence file uploaded. I can hear music. I will probably upgrade the NAA although it’s fairly recent.
CUDA offload is working but do I need to upgrade the drivers? Current CUDA version is 13.2. Driver 595.84.
One slight problem was that my interface was set to br0 which caused hqplayerd to crash. Changed it to auto and that fixed it.
I can see the new modulators. AHM7EC4B or AHM5EC4B?
IIRC, driver versions >=580 are supported for CUDA 13.4 used in latest v6 release. But I generally I would recommend to keep Nvidia drivers up to date.
I would start with AHM7EC4B, but depends on your amplifier too. If you have a class-D or tube amplifier, I would also try out AHM5EC4B.
Class D to valve. That’s seems to cover both ends of the spectrum. I am using Class D active loudspeakers so have started with AHM7EC4B. Processing load is slightly higher compared with my V5 setup at around 1.9 but still very comfortable for DSD1024.
The new UPNP stack seems to work well but the achilles heel is usually gapless playback. There was some slight stuttering between tracks but nothing to get excited about.
For track transition there are two approaches. The “stream as you” go and “freewheel”. Latter fetches the track to RAM at full speed, which may have CPU and network load implications. Although HQPlayer flags NAA traffic as high priority, so that doesn’t easily get affected if the networking infra supports traffic priorities properly.
So I would recommend to try both approaches. For gapless, also make sure your control point has gapless enabled for HQPlayer and you can check from HQPlayer that the upcoming track actually appears on the queue by latest 10 seconds before end of the current track. If it doesn’t, the control point is not sending the next track in time for gapless.