Hello community,
i am on the unstable branch, using nvidia only, and enabled wayland with the latest gpu drivers on Gnome.
I cannot say what it is exactly.
Verbose log of pipewire is normal.
Firefox also beside the known errors of mp4parse.
Symptom:
- Video plays normal.
- Randomly the sound softly goes down, but very fast
- After ~1sec it plays further as normal
- Similar to skipping to a new start point of a clip where it starts, but while playing constantly.
Results:
- Firefox in Youtube:
- Firefox in Invidious:
- Download and watch with mpv:
ok well now I have 2 audio devices:
- Internal
AMD FCH Azalia
over jack connection.
- External
NVIDIA GP107GL High Definition Audio
over Display Port
I switched off ASPM with TLP.
So my conclusion is, that it is a problem with
nvidia -> pipewire -> displayport
But I cannot narrow it down any further. I didn’t experience the problem on xorg.
Does anyone have an idea or found something that explain that behavior?
Meanwhile I will use the internal audio device.
Thank you for reading and suggestions.
Zesko
2
Wayland and Nvidia do not work together, I guess.
Maybe you can report the problem to Nvidia developers if they can support pipewire on Wayland.
ok I have maybe the root problem…
I run pipewire with triple -v
now and the journal reveals me this when a dropout happens:
Dez 09 14:52:43 pipewire[13524]: spa.alsa: 0x55ff47134f88: early wakeup 1021 1024
Dez 09 14:52:43 pipewire[13524]: spa.alsa: front:0: follower:0 match:0 rate:1,001412 bw:0,128000 thr:1024 del:1024 target:1024 err:0,000000 max:512,000000
Dez 09 14:52:43 pipewire[13524]: spa.alsa: hdmi:1,1: follower:0 match:0 rate:1,000016 bw:0,128000 thr:1024 del:1024 target:1024 err:0,000000 max:512,000000
Dez 09 14:52:43 pipewire[13524]: spa.alsa: 0x55ff47134f88: early wakeup 1023 1024
Dez 09 14:52:44 pipewire[13524]: spa.alsa: 0x55ff47134f88: early wakeup 1022 1024
Dez 09 14:52:44 pipewire[13524]: spa.alsa: 0x55ff47134f88: early wakeup 1022 1024
Dez 09 14:52:44 pipewire[13524]: spa.alsa: front:0: follower:0 match:0 rate:0,999993 bw:0,128000 thr:1024 del:1023 target:1024 err:-1,000000 max:512,000000
Dez 09 14:52:44 pipewire[13524]: spa.alsa: 0x55ff47134f88: early wakeup 1023 1024
Looks like a slight resynchronization of audio output…
No, it seems to work good enough at this stage even with nvidia. But well I am just testing at the moment.
Guess the problem is not the nvidia driver, since the audio device is running with snd_hda_intel
.
Puh… no idea what the actual problem was, but it seems again “power saving” *urgh*
I comment out:
File: /etc/modprobe.d/mhwd-gpu.conf
#options nvidia "NVreg_DynamicPowerManagement=0x02"
I added snd_hda_intel
and the device 01:00.1
to the deny list:
File: /etc/tlp.conf
RUNTIME_PM_BLACKLIST="01:00.1"
RUNTIME_PM_DRIVER_BLACKLIST="amdgpu mei_me nouveau nvidia pcieport radeon snd_hda_intel"
My guess is that it was the nvidia option in modprobe.
Now it works without interruption. At least I don’t hear it anymore.
No, it happens not that often anymore, but happens noticeably.