Schedule and location for the workshops is here. If you are interested in any of the workshops, please write your name in the list of interested people present in each workshop description.
Brief discussion about VKMS development, new ideas, user-space requirements, test issues.
Interested people:
The X server's GLX Vendor-Neutral Dispatch (GLXVND) layer provides a way for each X screen to use a different GLX implementation, but GLXVND is still limited to only one GLX implementation for each screen. On multi-GPU systems (such as laptops with both integrated and discrete GPUs), it would be useful to be able to run the desktop on one GPU and driver, but to allow some applications to run on a different GPU and driver, all within the same screen.
One possible solution is to extend the GLXVND interface to allow multiple GLX implementations on a screen, and then select between them on a per-client basis. There are some functional limitations to that by itself, though, such as mixing GLX and X11 rendering or accessing native X resources such as pixmaps. Other options such as using separate screens or even servers might be able to avoid those limits, but could have their own issues with complexity, scope, or compatibility.
I would like to discuss design ideas to try to figure out the best way to make this work.
Interested people:
Note: Should be scheduled after Emil's talk on gpu selection.
Interested people:
Note: Should be scheduled after the testing related talks by the interested people here.
Discussions of new kernel ABI that will be required to implement Vulkan timeline semaphores.