GPUSelect: Don't read depth on every draw

If the draw uses the same id as the previous draw, there is no need to read the depth
buffer at this point, avoiding a CPU-GPU sync bubble.

Fixes T62511 Selection is significantly slower in production scenes.
With this patch glReadPixels is not the bottleneck. Regular drawing Is still very
slow so I would suggest fixing the regular drawing first before trying to
improve the selection algorithm.

1 files changed, 7 insertions, 0 deletions

diff --git a/source/blender/gpu/intern/gpu_select_pick.c b/source/blender/gpu/intern/gpu_select_pick.c
index 6c3e05912b0..ffd4cd6efbc 100644
--- a/source/blender/gpu/intern/gpu_select_pick.c
+++ b/source/blender/gpu/intern/gpu_select_pick.c
@@ -479,7 +479,14 @@ static void gpu_select_load_id_pass_nearest(const DepthBufCache *rect_prev,
 bool gpu_select_pick_load_id(uint id)
 {
   GPUPickState *ps = &g_pick_state;
+
   if (ps->gl.is_init) {
+    if (id == ps->gl.prev_id) {
+      /* No need to read if we are still drawing for the same id since
+       * all these depths will be merged / deduplicated in the end. */
+      return true;
+    }
+
     const uint rect_len = ps->src.rect_len;
     glReadPixels(UNPACK4(ps->gl.clip_readpixels),
                  GL_DEPTH_COMPONENT,