Open jkms opened 2 years ago
I had this solution from this answer
In Short: Run this command
flatpak override --user --device=dri --env=__NV_PRIME_RENDER_OFFLOAD=1 --env=__VK_LAYER_NV_optimus=NVIDIA_only --env=__GLX_VENDOR_LIBRARY_NAME=nvidia org.blender.Blender
However, Cuda did not function for me :(
When I launch Blender via Flatpak on my nVidia Optimus laptop Blender correctly detects my nVidia card and it's CUDA abilities, but it doesn't actually use the nVidia GPU for rendering the viewport (notice that there's nothing in the output of
nvidia-smi
in the first screenshot):If I try to manually start blender with the Optimus/Prime environment variables:
However, when I start the non-flatpak version, it works correctly:
Flatpak
Like I said, blender flatpak works, and it sees my nVidia card:
Notice the GPU usage on my radeon iGPU vs the nVidia GPU.
Non-flatpak
However, when running the non-flatpak version with the Optimus/Prime env variables, I get buttery smooth performance, and the viewport rendering is offloaded to my nVidia card:
Is this a bug? or missing how to launch the flatpak correctly so that the viewport will use my nVidia GPU?