1dd7bb17c7
llvmpipe: Optimize blend swizzles by using bitmasks instead of shuffles for ubytes.
2009-08-29 09:21:23 +01:00
a77084ea4b
llvmpipe: Write data to a tsv file for posterior analysis.
2009-08-29 09:21:23 +01:00
f081bacc25
llvmpipe: Move p_build_context to lp_bld_type.h
...
As it will be shared with more modules.
2009-08-29 09:21:23 +01:00
696f7f2be5
util: New file dedicated to dump state in human/machine readable format.
2009-08-29 09:21:23 +01:00
5940ba26f7
llvmpipe: Don't use llvm.readcyclecounter.
...
LLVM during optimization reorders the rdtsc instructions next to each
other, which makes it pretty useless.
2009-08-29 09:21:22 +01:00
64611e086d
llvmpipe: Separate constant building.
2009-08-29 09:21:22 +01:00
c87fab0008
llvmpipe: Move type support functions into a separate file.
2009-08-29 09:21:22 +01:00
b94e22e4be
llvmpipe: Measure the number of cycles taken for blending.
2009-08-29 09:21:22 +01:00
5778970f15
llvmpipe: Minor comments / debug mnemonics.
2009-08-29 09:21:22 +01:00
02eaa32d83
llvmpipe: Optimize away min/max with equal operands.
2009-08-29 09:21:22 +01:00
9e92244440
llvmpipe: Improve generated 8bit blending code.
2009-08-29 09:21:22 +01:00
ede73258a7
llvmpipe: Get blending of normalized 8bit unsigned integers working.
2009-08-29 09:21:22 +01:00
a6622e6c54
llvmpipe: Specialize arithmetic operations.
2009-08-29 09:21:22 +01:00
272dadbe4e
llvmpipe: Introduce a custom typing system.
...
Straightforward representation of floating-point/fixed-point/integer,
normalized/scaled, signed/unsigned SIMD vector types.
2009-08-29 09:21:22 +01:00
f478b6fe76
llvmpipe: Avoid C99-isms.
2009-08-29 09:21:21 +01:00
fe90e46dc9
llvmpipe: Factor out intrisic calling code into a reusable function.
2009-08-29 09:21:21 +01:00
7ace0b1f4a
llvmpipe: Cleanup test programs.
2009-08-29 09:21:21 +01:00
7d043162c5
llvmpipe: Blending.
...
The code
2009-08-29 09:21:21 +01:00
d2cf3e8dfd
llvmpipe: Arithmetic helpers.
2009-08-29 09:21:21 +01:00
fa3514a57e
llvmpipe: fastpath for interpolated z16 less depthtesting
...
Because this is interpolated (ie. early) depth, we can build in an
assumption about the quads emitted by triangle setup, ie that they
are actually linear spans. Interpolate z over those spans in z16
format to save on math & conversion.
2009-08-29 09:21:21 +01:00
38a1479fe1
llvmpipe: cope with nr_cbufs == 0
...
Disable blend code when no color buffer
2009-08-29 09:21:21 +01:00
4639f8aad5
llvmpipe: Translate logicops.
2009-08-29 09:21:21 +01:00
833323b8d0
llvmpipe: Separate pixel packing/unpacking from loading/storing.
2009-08-29 09:21:21 +01:00
c4903ee27f
xlib: Integrate with llvmpipe.
...
Actually, the makefile/sconscript changes were already commit by mistake.
2009-08-29 09:21:21 +01:00
b060771612
llvmpipe: Adjust for pixel layout changes.
2009-08-29 09:21:20 +01:00
42b15b26fd
util: Make pixel format layout more meaningful.
...
Namelly, explicitly distinguish formats coded by arithmetic manipulation,
from formats layed out in array.
2009-08-29 09:21:20 +01:00
a910c7d820
util: Don't commit u_format_table.c
...
At least not until stabilizes, to reduce noise.
2009-08-29 09:21:20 +01:00
9a519ec4d1
llvmpipe: Fix typo in copyright.
2009-08-29 09:21:20 +01:00
3ef44746fe
llvmpipe: Add test cases for pixels.
...
Actually some of this test cases need more testing...
2009-08-29 09:21:20 +01:00
899bee2788
util: Fix r5g6b5 description.
2009-08-29 09:21:19 +01:00
3ace63c00f
llvmpipe: Update SConscript.
2009-08-29 09:21:19 +01:00
b023c1dc68
llvmpipe: Simple shell script to aid port softpipe changes to llvmpipe.
2009-08-29 09:21:19 +01:00
05bfd93c3f
llvmpipe: move all depth/stencil/alpha pixel processing into one stage
2009-08-29 09:21:19 +01:00
b8c437f20f
llvmpipe: fix error in scissor state dependencies
2009-08-29 09:21:18 +01:00
c5375da1dc
llvmpipe: cleanup framebuffer state routine slightly
2009-08-29 09:21:18 +01:00
90caf1a3f3
llvmpipe: move all color-combine code into lp_quad_blend.c
...
Consolidate the read-modify-write color combining code from
the blend, colormask and output stages.
2009-08-29 09:21:18 +01:00
54c6ef144f
llvmpipe: fix typo
2009-08-29 09:21:18 +01:00
3c71a19016
llvmpipe: example fastpaths in blending
2009-08-29 09:21:18 +01:00
80eb276630
llvmpipe: actually pass >1 quad from triangle routine
...
First attempt
2009-08-29 09:21:18 +01:00
bdbb4beb21
llvmpipe: expand quad pipeline to process >1 quad at a time
...
This is part one -- we still only pass a single quad down, but
the code can now cope with more. The quads must all be from the same
tile.
2009-08-29 09:21:18 +01:00
4486012245
llvmpipe: rip out old mulithread support
2009-08-29 09:21:18 +01:00
2a7c39a211
llvmpipe: avoid flushing depth buffer cache on swapbuffers
...
There's no need to push out depth buffer contents on swapbuffers.
Note that this change doesn't throw away depth buffer changes, it simply
holds them in the cache over calls to swapbuffers. The hope is
that swapbuffers will be followed by a clear() which means in that case
we won't have to write the changes out.
2009-08-29 09:21:17 +01:00
3adca96114
llvmpipe: also shortcircuit non-texture tile lookups
2009-08-29 09:21:17 +01:00
2301314e7c
llvmpipe: shortcircuit repeated lookups of the same tile
...
The lp_tile_cache is often called repeatedly to look up the same
tile. Add a cache (to the cache) of the single tile most recently
retreived and make a quick inline check to see if this matches the
subsequent request.
Add a tile_address bitfield struct to make this check easier.
2009-08-29 09:21:17 +01:00
df1823ec5b
llvmpipe: remove unused vars in lp_setup.c
2009-08-29 09:21:17 +01:00
08811032c2
llvmpipe: use bitwise logic to setup quad masks in lp_setup
2009-08-29 09:21:17 +01:00
0aa24fa36f
llvmpipe: simplify flush_spans
...
No loss of performance, but simpler code.
2009-08-29 09:21:17 +01:00
0f55a95b2f
llvmpipe: make some small steps to flush texture cache less frequently
...
No performance gain yet, but the code is a bit cleaner.
2009-08-29 09:21:16 +01:00
3121484a8b
llvmpipe: remove backwards dependency from tilecache to llvmpipe
...
The tile cache is a utility, it shouldn't know anything about the
entity which is making use of it (ie llvmpipe).
Remove llvmpipe parameter to all the tilecache function calls, and
also remove the need to keep a llvmpipe pointer in the sampler structs.
2009-08-29 09:21:16 +01:00
c9a5930fe4
llvmpipe: Rename preprocessor symbols too.
2009-08-29 09:21:16 +01:00
José Fonseca