080c40ab32
Merge remote branch 'origin/master' into lp-binning
...
Conflicts:
src/gallium/auxiliary/util/u_surface.c
src/gallium/drivers/llvmpipe/Makefile
src/gallium/drivers/llvmpipe/SConscript
src/gallium/drivers/llvmpipe/lp_bld_arit.c
src/gallium/drivers/llvmpipe/lp_bld_flow.c
src/gallium/drivers/llvmpipe/lp_bld_interp.c
src/gallium/drivers/llvmpipe/lp_clear.c
src/gallium/drivers/llvmpipe/lp_context.c
src/gallium/drivers/llvmpipe/lp_context.h
src/gallium/drivers/llvmpipe/lp_draw_arrays.c
src/gallium/drivers/llvmpipe/lp_jit.c
src/gallium/drivers/llvmpipe/lp_jit.h
src/gallium/drivers/llvmpipe/lp_prim_vbuf.c
src/gallium/drivers/llvmpipe/lp_setup.c
src/gallium/drivers/llvmpipe/lp_setup_point.c
src/gallium/drivers/llvmpipe/lp_state.h
src/gallium/drivers/llvmpipe/lp_state_blend.c
src/gallium/drivers/llvmpipe/lp_state_derived.c
src/gallium/drivers/llvmpipe/lp_state_fs.c
src/gallium/drivers/llvmpipe/lp_state_sampler.c
src/gallium/drivers/llvmpipe/lp_state_surface.c
src/gallium/drivers/llvmpipe/lp_tex_cache.c
src/gallium/drivers/llvmpipe/lp_tex_cache.h
src/gallium/drivers/llvmpipe/lp_tex_sample.h
src/gallium/drivers/llvmpipe/lp_tile_cache.c
2010-01-08 15:42:57 +00:00
c9240c4c8f
llvmpipe: remove dead code
2010-01-06 17:00:26 +00:00
808170a0ff
llvmpipe: replace INT_MIN/2 with INT_MIN
...
Since changing the in/out test we can just use INT_MIN to be sure the
comparison against the step values always passes.
2009-12-17 09:01:01 -07:00
b9d33db0a4
llvmpipe: improve the in/out test a little
...
Instead of:
s = c + step
m = s > 0
Do:
m = step > c (with negated c)
2009-12-17 08:17:04 -07:00
ab94381930
llvmpipe: do final the pixel in/out triangle test in the fragment shader
...
The test to determine which of the pixels in a 2x2 quad is now done in
the fragment shader rather than in the calling C code. This is a little
faster but there's a few more things to do.
Note that the step[] array elements are in a different order now. Rather
than being in row-major order for the 4x4 grid, they're in "quad-major"
order. The setup of the step arrays is a little more complicated now.
So is the course/intermediate tile test code, but some lookup tables
help with that.
Next steps:
- early-cull 2x2 quads which are totally outside the triangle.
- skip the in/out test for fully contained quads
- make the in/out comparison code tighter/faster.
2009-12-16 16:10:05 -07:00
cdbcd96fdf
llvmpipe: tighten up an assertion
2009-12-15 15:39:48 -07:00
663750d556
llvmpipe: rename bins to scene
...
It was pretty confusing having an entity named "bin" and another named
"bins", not least because sometimes there was a need to talk about >1
of the "bins" objects, which couldn't be pluralized any further...
Scene is a term used in a bunch of places to talk about what a binner
operates on, so it's a decent choice here.
2009-12-13 18:17:25 +00:00
4b70af918d
llvmpipe: added lp_rast_fence() bin function
2009-12-11 17:57:45 -07:00
9323740738
llvmpipe: added lp_rast_get_num_threads()
2009-12-11 17:45:52 -07:00
92dc0f92b0
llvmpipe: implement lp_rast_load_color()
2009-12-11 15:00:28 -07:00
2bce5c195f
llvmpipe: checkpoint: more thread/queuing changes
...
Now mapping/unmapping the framebuffer is done by a rasteizer thread
rather than the main calling thread.
2009-12-11 14:52:46 -07:00
de31b0e60c
llvmpipe: remove unused lp_rasterizer::fb field
2009-12-11 11:56:22 -07:00
156eabbaf9
llvmpipe: improve framebuffer/surface code
2009-12-11 11:46:23 -07:00
205da96fc6
llvmpipe: remove unused fb parameter
2009-12-11 11:25:22 -07:00
4e67f10331
llvmpipe: minor comment fix
2009-12-11 11:16:48 -07:00
9509f73c21
llvmpipe: checkpoint: use empty/full bin queues
2009-12-10 14:56:30 -07:00
96689d451a
llvmpipe: added some debug/info code
2009-12-09 16:32:32 -07:00
3bee8c2e7c
llvmpipe: use the empty_bins queue now
2009-12-09 16:06:04 -07:00
d7dbc66636
llvmpipe: checkpoint: begin plugging in bin queue code
2009-12-09 14:53:33 -07:00
7f457acabc
llvmpipe: Use number of CPUs as default number of threads.
...
Also bump MAX_THREADS to 8.
2009-12-08 08:02:49 +00:00
73e13c33fd
llvmpipe: remove some left-over debug code
2009-12-07 18:18:37 -07:00
aab1ceceec
llvmpipe: implement threaded rasterization
...
The LP_NUM_THREADS env var controls how many threads are created.
The default (and max) is 4, for now.
If LP_NUM_THREADS = 0, threading is not used.
2009-12-07 18:04:54 -07:00
3a06c113c7
llvmpipe: repartition lp_rasterizer state for threading
...
Some of the state is per-thread. Put that state in new lp_rasterizer_task
struct.
2009-12-07 18:04:54 -07:00
cdaea049c9
llvmpipe: use bin iteration functions when rasterizing bins
2009-12-07 18:04:54 -07:00
8a23105fa0
llvmpipe: still more bin code reorganization
...
Move tiles_x,y fields from setup state into bin state.
Move more bin-adding commands into lp_bin.[ch].
2009-12-04 15:59:25 -07:00
01b1900084
llvmpipe: reorganization of binning data structions and funtions
...
New lp_bins struct contains all bin information.
More move bin-related code into lp_bin.[ch]
Use new/updated bin-access functions to hide implementation details.
The result is more/cleaner separation between the setup and rast components.
This will make double-buffering of the bins easier, etc.
2009-12-04 15:31:09 -07:00
b533b56750
llvmpipe: move lp_rasterize_bin() into lp_rast.c
...
First step of moving bin rasterization/execution code out of lp_setup.c
2009-12-04 14:47:46 -07:00
a08d630216
llvmpipe: use LP_DBG() macro everywhere
2009-12-04 14:11:25 -07:00
b1659b9213
llvmpipe: bin state-change commands
...
Previously, each triangle had a pointer to the state to use for shading.
Now we insert state-change commands into the bins. When we execute one
of those commands we just update a 'current state' pointer and use that
pointer when calling the jit shader.
When inserting state-change commands into a bin we check if the previous
command was also a state-change command and simply replace it. This
avoids accumulating useless/redundant state-change commands.
2009-12-04 11:50:43 -07:00
e2f4634456
llvmpipe: checkpoint some initial state binning code
2009-12-03 16:05:12 -07:00
ffd0759973
llvmpipe: comments and a stub for lp_rast_set_state()
2009-12-03 14:57:44 -07:00
51663f0506
llvmpipe: quick & dirty implementation of lp_rast_store_zstencil()
...
This allows us to do a glReadPixels(GL_DEPTH_COMPONENT) to see what's
in the depth buffer to help debugging.
2009-12-03 11:41:48 -07:00
866e6856d3
llvmpipe: execute shaders on 4x4 blocks instead of 8x2
...
This matches the convention used by the recursive rasterizer.
Also fixed assorted typos, comments, etc.
Now tri-z.c, gears.c, etc look basically right but there's still some
cracks in triangle rasterization.
2009-12-02 15:13:47 -07:00
7505510c7b
llvmpipe: add a bunch of comments
2009-11-30 14:02:01 -07:00
280c8acc43
llvmpipe: fix the worst of the depth regressions since switch to 4x4
2009-10-22 17:21:37 +01:00
694f05ac18
llvmpipe: remove one of two definitions of TILESIZE
2009-10-21 15:21:11 +01:00
7b116e13a2
llvmpipe: pass mask as a linear encoding of the 4x4 block
2009-10-20 03:17:17 +01:00
7670628061
llvmpipe: precalculate some offsets
2009-10-20 02:46:00 +01:00
5b07d4de38
llvmpipe: remove a leftover 8x2 usage
2009-10-19 17:10:48 +01:00
b0828b0adc
llvmpipe: calculate masks in format desired by shader
...
Also remove branches calculating masks for quads.
2009-10-19 16:41:27 +01:00
61f3eeb640
llvmpipe: Use framebuffer coords consistently.
2009-10-09 19:16:36 +01:00
8c34c86d19
llvmpipe: Undo debug override.
2009-10-09 16:23:00 +01:00
05131f7502
llvmpipe: properly clip tile writes
2009-10-09 16:20:56 +01:00
c2e926b72d
llvmpipe: Pass framebuffer coords to shader.
2009-10-09 16:18:38 +01:00
82ec7f018d
llvmpipe: correct binning maths for iterating over whole tiles
2009-10-09 16:07:58 +01:00
8599969582
llvmpipe: Get jit_context/jit_function across the rasterizer.
2009-10-09 15:53:53 +01:00
e0e2008f1d
llvmpipe: trivial/clear works
2009-10-09 14:29:25 +01:00
295aea0489
llvmpipe: more debug
2009-10-09 14:07:40 +01:00
4e1334ced6
llvmpipe: debug, crash fixes
2009-10-09 14:02:39 +01:00
4cdd10cb4b
llvmpipe: use union lp_cmd_rast_arg directly, rather than through a pointer
...
The union itself consists of pointers. We don't need to be passing
pointer to pointers.
2009-10-09 11:29:01 +01:00
José Fonseca
Keith Whitwell
Brian Paul