nouveau : avoid vertex_size / 4 for each vertex ops
1 file changed