PRIME GPU Offloading Improvement For GLXVND Merged For X.Org Server 1.21

[phoronix]

Phoronix: PRIME GPU Offloading Improvement For GLXVND Merged For X.Org Server 1.21

Work by NVIDIA to provide separate per-client vendor mappings for GLXVND were merged to X.Org Server 1.21 Git as another step towards improving the PRIME GPU offloading support when multiple GPU drivers are at play...

http://www.phoronix.com/scan.php?page=news_item&px=GLXVND-Offload-Improve-1.21

[FireBurn]

As I always write with these articles, this is for nVidia only, PRIME works great with radeon & amdgpu using Mesa

[polarathene]

I suppose PRIME doesn't provide a way to expose a bridge from a linux VM guest to the host GPU? Could something like the Intel vGPU support be based off that or nope?

Can you offload to nvidia when using Wayland via this support and if so since it seems to mention X.org server, would it only work for apps running via XWayland?

[9Strike]

Does this mean we can simply offload to a secondary (laptop) gpu with PRIME=1 like with amd gpus?

[kmare]

Is this what is needed to replace bumblebee and finally be able to shutdown the Nvidia card cleanly in order to save battery life?

[MastaG]

So I was testing gnome 3.32 (fedora 30) with nvidia's drm so I could use wayland on eglstreams.
Everything works fine silk as butter, however Xwayland falls back to mesa softpipe.
Will this fix it so Xwayland can use nvidia's opengl driver now?

[uid313]

Maybe for X.Org Server 1.21 they can deprecate the x11-apps package or at least make it optional.

[andre30correia]

maybe dynamic official support finally land with intel nvidia laptops, without tricks or switch sessions

[kmare]

maybe dynamic official support finally land with intel nvidia laptops, without tricks or switch sessions

That's what I hope for too. At least it seems like that's what they're working on, based on this comment:

https://devtalk.nvidia.com/default/topic/957981/linux/prime-render-offloading-on-nvidia-optimus/post/5276481/#5276481