[MPlayer-dev-eng] Update to vo_direct3d - saves ~2% CPU time

[Georgi Petrov]

Thanks Jim, I'll take a look at your patch. Everyone - please note
that I may no be home this weekend, so I may not answer to your mails
in the following two days.

I had plans to modify myself big parts of the drawing code, because
it's far from optimal right now. Those include the surface locking,
scene begin/end, the needless stretchrects in draw_slice() and many
other things. I'll see what you have already done when I can.


>> I've optimized away some begin_scene and end_scene
>> combos as they are only needed for actual rendering, not the locking.

Are you sure? I understood that they are needed exactly when locking...

>From useless MSDN:

There should be at most one
IDirect3DDevice9::BeginScene/IDirect3DDevice9::EndScene pair between
any successive calls to present (either IDirect3DDevice9::Present or
IDirect3DSwapChain9::Present). IDirect3DDevice9::BeginScene should be
called once before any rendering is performed, and
IDirect3DDevice9::EndScene should be called once after all rendering
for a frame has been submitted to the runtime. To enable maximal
parallelism between the CPU and the graphics accelerator, it is
advantageous to call IDirect3DDevice9::EndScene as far ahead of
calling present as possible.

I understand that "rendering" means locking and stretchrect. See the
last sentance: it is advantageous to call IDirect3DDevice9::EndScene
as far ahead of calling present as possible. This means that EndScene
should be called when the rendering is done (stretchrect), so we can
execute Present afterward.