You are not logged in.
Pages: 1
quick rundown:
nvidia works 100% fine on xorg.
when I use kde's wayland session, it uses my amd renoir integrated graphics.
nvidia card is using optimus.
nvidia-smi shows that xorg is using the card (sddm) -- as intended.
if i disable sddm.service, reboot, and run startplasma-wayland, it doesnt work. however it shows no processes using the nvidia card (confirms sddm is using the xorg process above)
wayland runs on tty2 while sddm is running on tty1 (sddm's fault).
Last edited by thealphareturns090123894 (2022-03-03 23:08:41)
Offline
All of that is logically speaking normal, from a plain technical standpoint you "should" be able to use prime-run to offload things, however I'm highly sceptical on how well this generally works on Wayland.
FWIW base requirement for any wayland support on nvidia is enabling early KMS: https://wiki.archlinux.org/title/NVIDIA … de_setting
Offline
yes i did that. modules are in mkinitcpio.conf (+ i915) and the nvidia-drm.modeset=1 is in the grub kernel parameters
Last edited by thealphareturns090123894 (2022-03-04 01:47:28)
Offline
yes i did that. modules are in mkinitcpio.conf (+ i915)
What actual modules are enable in mkinitcpio? i915 is for intel graphics.
Offline
Presumingly the nvidia modules, and if xorg works correctly then the module situation should be fine.
Since there's no global server on wayland, even if this was working in some form, I'd not expect any process to land on the nvidia GPU by default anyway. While I'm again less than certain this even works (for any situation, I'm not sure it would work with a dedicated AMD card either), are you able to just run prime-run $application and see that process land on the nvidia card? Does
prime-run vulkaninfo #... glxinfo, eglinfo from mesa-utils/vulkan-tools respectively
work? There are a lot of logical gymnastics that have to happen to make this in general happen on wayland that are simply no problem (... but in turn for being easy there have relevant privacy/security concerns) on xorg like sharing application buffers willy nilly
Last edited by V1del (2022-03-04 10:18:19)
Offline
all the nvidia modules + i915 are in the mkinitcpio
prime-run vulcaninfo, glxinfom eglinfo all show nvidia
running prime-run startplasma-wayland in the next available tty blackscreens
Last edited by thealphareturns090123894 (2022-03-04 11:59:55)
Offline
I'd just not generally do that in any case, and this kind of bend over the entire nvidia image back over the iGPU is absolutely guaranteed to not yet work. Check whether you can run real single applications like actual games with prime-run. You really shouldn't don't want to target and run the entire session on the dedicated GPU for this usecase.
Last edited by V1del (2022-03-04 12:17:31)
Offline
Pages: 1