[Glamor] glamor: Changes to 'master'

[Michel Dänzer]

On Mit, 2012-07-11 at 00:59 -0700, Zhigang Gong wrote: 
> 
> commit 593bd206bf2794dc1450e8e7d20142b8d0ca074b
> Author: Zhigang Gong <zhigang.gong at linux.intel.com>
> Date:   Tue Jun 26 15:39:24 2012 +0800
> 
>     glamor_render: Don't allocate buffer for vbo each time.
>     
>     We can reuse the last one if the last one is big enough
>     to contain current vertext data. In the meantime, Use
>     MapBufferRange instead of MapBuffer.
>     
>     Testing shows, this patch brings some benefit for
>     aa10text/rgb10text. Not too much, but indeed faster.
>     
>     Signed-off-by: Zhigang Gong <zhigang.gong at linux.intel.com>

This commit decreases glyph throughput by about a factor of 4 with the
r600g driver on an RS880 laptop.

AFAICT the new code always maps the VBO from offset 0, which I'm not
sure can be expected to perform well. However, I've tried sliding the
mapping range over a larger VBO, and it doesn't seem to help. Maybe the
r600g driver doesn't handle this kind of usage very well yet.

Still, how would you feel about reverting this change for now?


OTOH the other changes in this push about double the glyph throughput
(actually surpassing EXA) on the same machine, nice!


-- 
Earthling Michel Dänzer           |                   http://www.amd.com
Libre software enthusiast         |          Debian, X and DRI developer