Archived

This topic is now archived and is closed to further replies.

glNoname3f

vertex sharing

Recommended Posts

If you want information about compiled vertex arrays, take a look at the GL_EXT_compiled_vertex_arrays extension specifications.

You will certainly increase performance if you have a unique vertex array in your 3D world, but you will not necessarily increase performance if you have more than one vertex array visible simultaneously in the viewing frustum.

Share this post


Link to post
Share on other sites
quote:
Original post by vincoof

You will certainly increase performance if you have a unique vertex array in your 3D world, but you will not necessarily increase performance if you have more than one vertex array visible simultaneously in the viewing frustum.


The problem is that I have one interleaved array, use glLockArrays function, draw array with glDrawElements, tristrips.
I use several glDrawElements and there is one shared row every time, but performance is like without locking.

Share this post


Link to post
Share on other sites
that isn''t surprising. Look into optimizing for the verex cache (which is usually about 10 verticies long) this will reward you with significantly higher poly throughputs assuming that your tris/sec rate is the bottle neck (which it often isn''t).

Share this post


Link to post
Share on other sites
quote:
Original post by RipTorn
that isn''t surprising. Look into optimizing for the verex cache (which is usually about 10 verticies long) this will reward you with significantly higher poly throughputs assuming that your tris/sec rate is the bottle neck (which it often isn''t).


Thanks, but I failed again. About vertex cache: does it mean that I should process less data with a single glDrawElements (10 verticies or less)? Maybe I shouldn''t use display list?
How does this vertex cache works and when?

Share this post


Link to post
Share on other sites