Skip to content

vbo/dlist: another set of CPU optimizations and cleanups

This series started as a suggestion from Marek to use a single bo for both the vertices and the indices.

I used this opportunity to cleanup the vbo display list code.

The end result is less draw calls (because successive draw calls using the same bo can be merged in a single draw), less code and better performance in viewperf13 snx (about 10%).

Merge request reports