sungmg/mutter-performance-source
1
0
Fork 0
Code Activity

[pango-display-list] Use the Cogl journal for short runs of text

Browse source 
For small runs of text like icon labels, we can get better performance
going through the Cogl journal since text may then be batched together
with other geometry.

For larger runs of text though we still use VBOs since the cost of logging
the quads becomes too expensive, including the software transform which
isn't at all optimized at this point.  VBOs also have the further advantage
of avoiding repeated validation of vertices by the driver and repeated
mapping of data into the GPU so long as the text doesn't change.

Currently the threshold is 100 vertices/25 quads. This number was plucked
out of thin air and should be tuned later.

With this change I see ~180% fps improvment for test-text. (x61s + i965 +
Mesa 7.6-devel)
This commit is contained in:
Robert Bragg 2009-06-17 23:35:49 +01:00
parent 6ac3b5a564
commit 6562f3224a
1 changed files with 33 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

33
clutter/pango/cogl-pango-display-list.c
View file

@ -244,6 +244,39 @@ _cogl_pango_display_list_render_texture (CoglHandle material,
cogl_material_set_color (material, &premult_color);
cogl_set_source (material);
/* For small runs of text like icon labels, we can get better performance
* going through the Cogl journal since text may then be batched together
* with other geometry. */
/* FIXME: 100 is a number I plucked out of thin air; it would be good
* to determine this empirically! */
if (node->d.texture.verts->len < 100)
{
int i;
for (i = 0; i < node->d.texture.verts->len; i += 4)
{
CoglPangoDisplayListVertex *v0 =
&g_array_index (node->d.texture.verts,
CoglPangoDisplayListVertex, i);
CoglPangoDisplayListVertex *v1 =
&g_array_index (node->d.texture.verts,
CoglPangoDisplayListVertex, i + 2);
cogl_rectangle_with_texture_coords (v0->x, v0->y, v1->x, v1->y,
v0->t_x, v0->t_y,
v1->t_x, v1->t_y);
}
return;
}
/* It's expensive to go through the Cogl journal for large runs
* of text in part because the journal transforms the quads in software
* to avoid changing the modelview matrix. So for larger runs of text
* we load the vertices into a VBO, and this has the added advantage
* that if the text doesn't change from frame to frame the VBO can
* be re-used avoiding the repeated cost of validating the data and
* mapping it into the GPU... */
if (node->d.texture.vertex_buffer == COGL_INVALID_HANDLE)
{
CoglHandle vb = cogl_vertex_buffer_new (node->d.texture.verts->len);