Maintainer has been absent for some time so kernel v6.11 and v6.12 isn't supported OOTB, to get it to work with kernel v6.11 you need to pull the fix from: !48
SavvyBeardedFish
Additionally you can try and force use amdgpu
rather than radeon, by setting the kernel flags:
radeon.cik_support=0 radeon.si_support=0 amdgpu.cik_support=1 amdgpu.si_support=1 amdgpu.dc=1
Device initalization failed according to the Xorg logs;
- Dump your firmware version
- Dump your kernel version
- Dump your kernel logs (
dmesg
orjournalctl -k
)
Running:
swaymsg for_window "[app_id=mpv] opacity 0.5"
Works as expected on my end, are you missing just executing for_window
?
Note, you can also add multiple rules in the same execution, e.g.
for_window {
[app_id=mpv] opacity 0.85
[app_id=LibreWolf] opacity 0.85
}
Also, note that app_id
of LibreWolf is capitalized in that manner.
You can get that information [app_id, shell etc] by running swaymsg -t get_tree
Feel like most people still do the scripting in Bash for portability reasons, and then just run Fish as the interactive shell
Nice, then you should be able to run vkcube
to verify whether your GPU is activated properly.
You can do several "iterations" here as well.
- Install Mangohud so you can visibly see if your GPU is activated correctly
- Run
mangohud vkcube-wayland
- Does it use your Nvidia GPU? - Run
mangohud vkcube
- Does it use your Nvidia GPU?
If Step 2 nor 3 shows your Nvidia GPU you can try and force it with:
mangohud vkcube-wayland --gpu_number 0
Start with the basics, do you see your Nvidia GPU pop up when using vulkaninfo --summary
?
If it doesn't pop up, verify that you have the correct vulkan ICD files in:
$ ls /usr/share/vulkan/icd.d/
There you should have nvidia_icd.json, nvidia_layers.json
.
If that's missing, you're missing the nvidia-utils part of the driver.
If they are there, but it still don't show in your vulkaninfo sumary, you could try to load the nvidia driver manually; modprobe nvidia
, also check the kernel logs journalctl -k
or dmesg
and search for nvidia
to see whether the driver got loaded correctly?
Breaking Linux every week or every other week? That's almost impressive!
If all nodes are connected through ethernet to each other (or at least one common node) you could go for OpenWRT's 'Dumb AP' setup as well
Edit: Already mentioned here; https://feditown.com/comment/1980836