llvmpipe: optimize the lp_setup_fence() function

Avoid emitting fences when not needed.
Speeds up glReadPixels quite a bit when reading image row by row.
This commit is contained in:
Brian Paul
2010-03-24 20:49:12 -06:00
parent d7ddb589f4
commit a9063cad0f

View File

@@ -318,16 +318,30 @@ lp_setup_fence( struct lp_setup_context *setup )
{
struct lp_scene *scene = lp_setup_get_current_scene(setup);
const unsigned rank = lp_scene_get_num_bins( scene ); /* xxx */
struct lp_fence *fence = lp_fence_create(rank);
struct lp_fence *fence;
LP_DBG(DEBUG_SETUP, "%s rank %u\n", __FUNCTION__, rank);
set_scene_state( setup, SETUP_ACTIVE );
if (setup->state == SETUP_FLUSHED) {
/* We're in a flushed state so there's nothing in the bins.
* No need to wait on a fence.
*/
fence = NULL;
}
else {
/* There's material in the bins. Emit the fence into the bins.
* When the rasterizer(s) find the fence, they'll signal on it.
*/
fence = lp_fence_create(rank);
/* insert the fence into all command bins */
lp_scene_bin_everywhere( scene,
lp_rast_fence,
lp_rast_arg_fence(fence) );
set_scene_state( setup, SETUP_ACTIVE );
/* insert the fence into all command bins */
lp_scene_bin_everywhere( scene,
lp_rast_fence,
lp_rast_arg_fence(fence) );
}
return (struct pipe_fence_handle *) fence;
}