07-19-2010, 10:36 AM
Yeah, that's line 836 in graphics_renderers.cpp. Probably only really needs to be done once, however it should only get called in the non-shader path.
Pages: 1 2
07-19-2010, 10:36 AM
12-23-2010, 09:32 AM
Have been profiling cardynamics changes lately and noticed that we are spending a lot of time in DRAWABLE_CONTAINER_HELPER::AccumulateSize::operator(). I wonder if we are missing an optimization opportunity here.
Code:
13.75 1.76 1.76 _mcount_private
12.11 3.31 1.55 16467397 0.00 0.00 void DRAWABLE_CONTAINER_HELPER::AccumulateSize::operator()<keyed_container<DRAWABLE> >(keyed_container<DRAWABLE> const&)
7.73 4.30 0.99 30452 0.00 0.00 void SCENENODE::Traverse<PTRVECTOR>(DRAWABLE_CONTAINER<PTRVECTOR>&, MATRIX4<float> const&)
3.75 4.78 0.48 16986 0.00 0.00 void AABB_SPACE_PARTITIONING_NODE<DRAWABLE*, 1u>::Query<FRUSTUM, std::vector<DRAWABLE*, std::allocator<DRAWABLE*> > >(FRUSTUM const&, std::vector<DRAWABLE*, std::allocator<DRAWABLE*> >&) const
2.58 5.11 0.33 1333989 0.00 0.00 RENDER_INPUT_SCENE::SelectTexturing(DRAWABLE&, GLSTATEMANAGER&)
2.34 5.41 0.30 1333964 0.00 0.00 RENDER_INPUT_SCENE::SelectTransformStart(DRAWABLE&, GLSTATEMANAGER&)12-23-2010, 09:08 PM
Good catch. It was happening because the scenenode was doing an early out if its drawlist was empty, and empty() was calling size() which is more expensive than I realized. Traverse() is so fast (it just seems slow because it's called so frequently) that it's actually better to skip the early-out check entirely.
12-24-2010, 10:42 AM
We need a wiki page on how to profile VDrift. If anyone would like to start one, they could probably use some of the information in this thread.
http://wiki.vdrift.net/VDrift_Documentat...ction=edit
http://wiki.vdrift.net/VDrift_Documentat...ction=edit
12-24-2010, 01:27 PM
OK, I put something really simple there.
Pages: 1 2