Age | Commit message (Collapse) | Author |
|
In particular, don't use the clamped lod to compute level + 1, or
lod in [-1, 0] range will actually interpolate with level 1.
This makes Mipfilter DCT pass 100%.
|
|
Also start axing the code duplication for scalar case. The olution is to
treat the scalar case specially in a few innermost functions, and leave
outer functions untouched.
|
|
It's quite a pain to remember the details after a while, and it is quite
likely we'll want to use this again, either for different polynomial
orders or different functions, so commit it here.
|
|
The advantage of range[-0.5, 0.5] is that it doesn't require floor (for
which intrinsics are only available in SSE4.1).
But the EXP opcode pretty much forces us to use floor, and there is a
good floor approximation around truncation available anyway.
This fixes EXP failures in VShader DCT.
|
|
|
|
A quick hack to get the right results, as there are many DCT tests
which use these opcodes to generate data to test other opcodes.
|
|
The cont_mask must be restored and exec mask recomputed in order to decide
whether to repeat the loop or not.
Unlike the continue mask, the break_mask must be preserved across loop
iterations.
Fixes several VShader DCT cases, and no regressions with glean.
|
|
|
|
|
|
|
|
|
|
Part of a fix for piglit trinity-fp1 test failure.
|
|
Note that this also requires X86 for llvm, if llvmpipe/draw_llvm works
on PPC then the condition should be extended to include && x86.
Signed-off-by: Török Edwin <edwintorok@gmail.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
|
|
To properly execute an instruction such as "ADD tmp, tmp.wzyx, foo;"
with SOA we (sometimes) need to put the results into temporaries before
writing the results to the destination register.
This patch fixes the ADD instruction but this needs to be done for
many more instructions.
Helps to fix piglit fp-long-alu test (fd.o bug 27989).
|
|
|
|
|
|
This lets us unbind a shader from the tgsi_exec_machine. Since
shaders aren't ref counted we need this to properly clean up when
deleting shaders elsewhere.
|
|
|
|
Concidering that we actually print FALSE when displaying the
option we should also accept that value.
|
|
These files are built with make and removed with make clean, so it does not
seem necessary to track them.
Looking at the Makefile, it seems that the two u_indices_* files are handled
similarly to u_format_srgb.c u_format_table.c and u_half.c, and these 3
files are already untracked and in .gitignore
Signed-off-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
|
|
This incomplete patch got commited by mistake.
This reverts commit 2142c769a4ebfe1a7c3facb036af8b75c5288616.
|
|
These files are built with make and removed with make clean, so it does not
seem necessary to track them.
Looking at the Makefile, it seems that the two u_indices_* files are handled
similarly to u_format_srgb.c u_format_table.c and u_half.c, and these 3
files are already untracked and in .gitignore
Signed-off-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
|
|
These files are built with make and removed with make clean, so it does not
seem necessary to track them.
Looking at the Makefile, it seems that the two u_indices_* files are handled
similarly to u_format_srgb.c u_format_table.c and u_half.c, and these 3
files are already untracked and in .gitignore
Signed-off-by: Xavier Chantry <chantry.xavier@gmail.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
|
|
This fixes edge flags for polygons and provoking vertex for
filled quads.
|
|
|
|
Now we have draw_elements() and draw_arrays() to be consistent with the
pipe_context drawing functions.
|
|
Builds on commit ddb0e18f6c5582d4d2cc59ffd16ad9c4639ed059 and fixes
regressions in glean clipFlat test.
We assume that Gallium drivers observe flatshade_first for all triangles
and that all the assorted per-triangle calls in the 'draw' module also
follow flatshade_first. Everything else builds on those rules.
Gallium does not use follow flatshade_first for GL quads, quad strips
and polygons; the "last" vertex is always the provoking vertex for those
prims. So now there are separate QUAD_FIRST_PV and QUAD_LAST_PV macros
in the draw primitive decomposition code instead of one QUAD macro.
|
|
|
|
|
|
|
|
Based on John Lawless' feedback.
|
|
|
|
|
|
See FDO bug #27887.
Signed-off-by: Marek Olšák <maraeo@gmail.com>
|
|
|
|
|
|
|
|
This is a better place than in u_rect.c
|
|
|
|
Disabled as it doesn't make VS/PSPrecision DCT happy, and it would
unnecessarily slow some cases where it is not needed.
|
|
|
|
|
|
|
|
|
|
This gives a ~30% shader optimization time improvement on blender.
Tested by comparing the dumped LLVM modules.
Current ordering:
time ~/llvm-git/obj/Release-Asserts/bin/opt l.bc -constprop -instcombine
-mem2reg -gvn -simplifycfg
real 0m1.126s
user 0m1.108s
sys 0m0.012s
With this patch:
time ~/llvm-git/obj/Release-Asserts/bin/opt l.bc -mem2reg -constprop -instcombine -gvn -simplifycfg
real 0m0.885s
user 0m0.880s
sys 0m0.000s
The overall improvement in blender is ~15%.
Blender without the patch takes 1m13s:
edwin 5934 87.6 11.5 729440 458296 pts/5 SLl+ 17:35 1:13 blender
Blender with the patch takes 1m3s:
edwin 5726 94.2 11.2 716424 446168 pts/5 SLl+ 17:32 1:03 blender
It is still slow with the patch, but better (most of the optimization time is
taken up by GVN, see LLVM PR7023).
Signed-off-by: Török Edwin <edwintorok@gmail.com>
Signed-off-by: José Fonseca <jfonseca@vmware.com>
|
|
supplied for tgsi translation.
|
|
PIPE_FORMAT_R10G10B10X2_USCALED, half floats, were not supported, so
just rely on u_format for (almost) universal format support.
|
|
Unneeded since we code generate the vertex fecthes.
|
|
|
|
It was only used for D3D's REP/END/BGNFOR/ENDFOR. D3D's aL register is
just like another address register now.
|