AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
53,238 Commits 1 Branch 0 Tags
d213d27f840598c19caa1a9fbf2e9a52fcbd6fcb
Commit Graph

16964 Commits

Author SHA1 Message Date
Tom Stellard b1e7bd7690 r600g: Fix segfault in r600_compute_global_transfer_map() 
This segfault was caused by commit
369e468889, however it is my fault for not
testing the patch while it was on the list.
 2012-10-16 14:39:16 +00:00
Tom Stellard a73c5d3f9d r600g: Fix build with --enable-opencl 2012-10-16 14:39:15 +00:00
Andreas Boll c5adfb21b3 r600g: drop useless switch statement 
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
 2012-10-15 20:34:02 +02:00
Andreas Boll 0ce21660c2 gallium/docs: update some distro information 
Reviewed-by: Brian Paul <brianp@vmware.com>
 2012-10-15 16:11:49 +02:00
Marek Olšák 023dae71ef r600g: emit the border color only when it's needed 
That depends on the texture wrap modes and filtering.
 2012-10-15 16:04:09 +02:00
Marek Olšák 33dda8f4fb r600g: cleanup create_sampler_state functions 
- stopped using util_color
- reformatted to occupy less characters per line.
- used memcpy for the border color
- used pipe_color_union in the state structure
 2012-10-15 16:04:09 +02:00
Brian Paul 1ec12c53ba util: added debug_print_transfer_flags() function 2012-10-15 07:49:14 -06:00
Brian Paul 88ecd0ddb9 svga: remove needless debug-mode linked list code 
LIST_DEL() always sets the prev/next pointers to NULL now.
 2012-10-15 07:49:14 -06:00
Chris Fester 3fffe8f7b7 util: null-out the node's prev/next pointers in list_del() 
Note: This is a candidate for the 9.0 branch.
Signed-off-by: Brian Paul <brianp@vmware.com>
 2012-10-15 07:49:14 -06:00
Christoph Bumiller 43e6c51aed nouveau: fix offset in nouveau_buffer_transfer_map 
Before 369e468889, the transfer was
initialized before the call to map and had the correct value already.
 2012-10-14 18:58:04 +02:00
Matt Turner fb85b204d3 u_format_s3tc.c: Don't call getenv() twice 
Reviewed-by: Brian Paul <brianp@vmware.com>
 2012-10-12 12:26:03 -07:00
José Fonseca bf2edc776b gallivm: Don't use llvm.x86.avx.max/min.ps.256 inadvertently. 
Could happen when CPU supports AVX, but LLVM doesn't.
 2012-10-12 18:52:28 +01:00
José Fonseca 9ccf91f9ef tgsi: Dump register number when dumping immediates. 
For example:

VERT
DCL IN[0]
DCL OUT[0], POSITION
DCL OUT[1], GENERIC[12]
DCL CONST[0..4]
DCL TEMP[0], LOCAL
DCL TEMP[1], LOCAL
IMM[0] UINT32 {4294967295, 0, 0, 0}
IMM[1] FLT32 {    0.0000,     1.0000,     0.0000,     0.0000}
  0: SEQ TEMP[0].x, CONST[3].xxxx, IMM[0].xxxx
  1: F2I TEMP[0].x, -TEMP[0]
  2: SEQ TEMP[1].x, CONST[4].xxxx, IMM[0].xxxx
  3: F2I TEMP[1].x, -TEMP[1]
  4: AND TEMP[0].x, TEMP[0].xxxx, TEMP[1].xxxx
  5: IF TEMP[0].xxxx :0
  6:   MOV TEMP[0], IMM[1].xyxy
  7: ELSE :0
  8:   MOV TEMP[0], IMM[1].yxxy
  9: ENDIF
 10: MOV OUT[1], TEMP[0]
 11: MOV OUT[0], IN[0]
 12: END

instead of

VERT
DCL IN[0]
DCL OUT[0], POSITION
DCL OUT[1], GENERIC[12]
DCL CONST[0..4]
DCL TEMP[0], LOCAL
DCL TEMP[1], LOCAL
IMM UINT32 {4294967295, 0, 0, 0}
IMM FLT32 {    0.0000,     1.0000,     0.0000,     0.0000}
  0: SEQ TEMP[0].x, CONST[3].xxxx, IMM[0].xxxx
  1: F2I TEMP[0].x, -TEMP[0]
  2: SEQ TEMP[1].x, CONST[4].xxxx, IMM[0].xxxx
  3: F2I TEMP[1].x, -TEMP[1]
  4: AND TEMP[0].x, TEMP[0].xxxx, TEMP[1].xxxx
  5: IF TEMP[0].xxxx :0
  6:   MOV TEMP[0], IMM[1].xyxy
  7: ELSE :0
  8:   MOV TEMP[0], IMM[1].yxxy
  9: ENDIF
 10: MOV OUT[1], TEMP[0]
 11: MOV OUT[0], IN[0]
 12: END
 2012-10-12 18:52:14 +01:00
Roland Scheidegger d366520e85 gallivm: fix rsqrt failures 
lp_build_rsqrt initially did not do any newton-raphson step. This meant that
precision was only ~11 bits, but this handled both input 0.0 and +infinity
correctly. It did not however handle input 1.0 accurately, and denormals
always generated infinity result.
Doing a newton-raphson step increased precision significantly (but notably
input 1.0 still doesn't give output 1.0), however this fails for inputs
0.0 and infinity (both result in NaNs).
Try to fix this up by using cmp/select but since this is all quite fishy
(and still doesn't handle denormals) disable for now. Note that even with
workarounds it should still have been faster since the fallback uses sqrt/div
(which both use the usually unpipelined and slow divider hw).
Also add some more test values to lp_test_arit and test lp_build_rcp() too while
there.

v2: based on José's feedback, avoid hacky infinity definition which doesn't
work with msvc (unfortunately using INFINITY won't cut it neither on non-c99
compilers) in lp_build_rsqrt, and while here fix up the input infinity case
too (it's disabled anyway). Only test infinity input case if we have c99,
and use float cast for calculating reference rsqrt value so we really get
what we expect.

Reviewed-by: José Fonseca <jfonseca@vmware.com>
 2012-10-12 18:51:18 +01:00
José Fonseca 2a4105cbc0 galahad: galahad_context_blit 
must unwrap.
 2012-10-12 18:38:05 +01:00
Marek Olšák 555c8d500a r600g: move shader structures into r600_shader.h 2012-10-12 19:00:30 +02:00
Marek Olšák 7997b3c97c r600g: implement MSAA resolving for 8-bit and 16-bit integer formats 
by changing the format to NORM.
 2012-10-12 15:23:27 +02:00
Brian Paul 743d859e62 util: fix broken pipe_get_tile_rgba() call 
Fix breakage from commit 369e468.
 2012-10-11 15:53:16 -06:00
Tom Stellard 4cc530f452 radeon/llvm: Fix build with LLVM 3.2 2012-10-11 21:33:00 +00:00
Tom Stellard dc54c49df9 clover: Fix build with LLVM 3.2 
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
 2012-10-11 21:32:54 +00:00
Tom Stellard c6b0132d1e clover: Don't link against libclangRewrite 
This library does not exist in LLVM 3.2 and libOpenCL.so links fine
without it on LLVM 3.1

Reviewed-by: Francisco Jerez <currojerez@riseup.net>
 2012-10-11 21:32:36 +00:00
Marek Olšák 7b01bc1e4c radeonsi: handle unhandled CAPs 2012-10-11 21:36:26 +02:00
Marek Olšák dd9274df4f radeonsi: fixup the return type of is_format_supported 2012-10-11 21:32:47 +02:00
Marek Olšák 8e3e4145ce radeonsi: remove unused local variables 2012-10-11 21:31:36 +02:00
Marek Olšák 47b7af6337 r600g: put user indices in the command stream for small index counts 
This improves performance a little bit if there are lots of small indexed
draw commands.
 2012-10-11 21:21:59 +02:00
Marek Olšák 0369fc9725 r600g: inline r600_translate_index_buffer 2012-10-11 21:21:34 +02:00
Marek Olšák 369e468889 gallium: unify transfer functions 
"get_transfer + transfer_map" becomes "transfer_map".
"transfer_unmap + transfer_destroy" becomes "transfer_unmap".

transfer_map must create and return the transfer object and transfer_unmap
must destroy it.

transfer_map is successful if the returned buffer pointer is not NULL.
If transfer_map fails, the pointer to the transfer object remains unchanged
(i.e. doesn't have to be NULL).

Acked-by: Brian Paul <brianp@vmware.com>
 2012-10-11 21:12:16 +02:00
Marek Olšák 9fe06f8815 softpipe: remove unused functions 
Reviewed-by: Brian Paul <brianp@vmware.com>
 2012-10-11 21:12:10 +02:00
Brian Paul 60a9390978 svga: don't use uninitialized framebuffer state 
Only the first 'nr_cbufs' color buffers in the pipe_framebuffer_state are
valid.  The rest of the color buffer pointers might be unitialized.
Fixes a regression in the piglit fbo-srgb-blit test since changes in the
gallium blitter code.

NOTE: This is a candidate for the 9.0 branch (just to be safe).

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
 2012-10-11 09:13:59 -06:00
John Kåre Alsaker 6c53ec1ef2 svga: Remove wierd code which forces non-sRGB formats. 
Signed-off-by: Brian Paul <brianp@vmware.com>
 2012-10-10 18:22:22 -06:00
John Kåre Alsaker 1a4aad11b0 svga: Add support for 16-bit per channel RGBA 
Signed-off-by: Brian Paul <brianp@vmware.com>
 2012-10-10 18:19:44 -06:00
Vincent Lejeune 5090ce42e4 radeon/llvm: use ceil intrinsic instead of llvm.AMDIL.round.posinf 
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
 2012-10-10 22:03:33 +02:00
Vincent Lejeune 9a6bb3f645 radeon/llvm: use floor intrinsic instead of llvm.AMDIL.floor 
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
 2012-10-10 22:03:20 +02:00
Vincent Lejeune bfdf26892c radeon/llvm: use llvm fabs intrinsic 
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
 2012-10-10 22:03:03 +02:00
Vincent Lejeune 8db11bc4ed radeon/llvm: use llvm intrinsic for flog2 
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
 2012-10-10 22:02:45 +02:00
Vincent Lejeune 23e11ac835 radeon/llvm: add support for cos/sin intrinsic 
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
 2012-10-10 22:02:28 +02:00
Vincent Lejeune 876b42663c radeon/llvm: add a pattern for fsqrt 
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
 2012-10-10 22:02:13 +02:00
Marek Olšák 87a34131c4 r600g: move SQ_GPR_RESOURCE_MGMT_1 into new config_state 
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:17:07 +02:00
Marek Olšák c5584e93b1 r600g: move DB_SHADER_CONTROL into db_misc_state 
Also update the register value in more appropriate places
than r600_update_derived_state.

Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:17:05 +02:00
Marek Olšák ae25b93245 r600g: emit PS_PARTIAL_FLUSH at the beginning of CS 
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:17:03 +02:00
Marek Olšák ef723613e0 r600g: atomize depth-stencil-alpha state 
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:17:01 +02:00
Marek Olšák 711f3bae9d r600g: atomize rasterizer state 
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:58 +02:00
Marek Olšák 9a683d1bd8 r600g: sort variables in r600_context 
Some variables have been removed from there too.

Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:55 +02:00
Marek Olšák 30bcc5538f r600g: initialize SQ_VTX_SEMANTIC_* in the start_cs command buffer 
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:49 +02:00
Marek Olšák 18a189188a r600g: atomize scissor state 
The workaround for R600 lacking VPORT_SCISSOR_ENABLE has also been simplified.

Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:45 +02:00
Marek Olšák ab075de53b r600g: atomize polygon offset state 
POLY_OFFSET_DB_FMT_CNTL is moved to the framebuffer state, because it only
depends on the zbuffer format.

Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:42 +02:00
Marek Olšák a50edc8ed8 r600g: atomize fetch shader 
The state object is actually a buffer, it's literally a buffer containing
the shader code.

Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:38 +02:00
Marek Olšák 8bf7044ec6 r600g: remove the dual_src_blend flag from the shader key 
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:35 +02:00
Marek Olšák faaba52aed r600g: atomize blend state 
This is not so trivial, because we disable blending if the dual src
blending is turned on and the number of color outputs is less than 2.
I decided to create 2 command buffers in the blend state object and just
switch between them when needed, because there are other states unrelated
to blending (like the color mask) and those shouldn't be changed
(the old code had it wrong).

Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:32 +02:00
Marek Olšák eb65fefa4b r600g: inline r600_atom_dirty 
Reviewed-by: Jerome Glisse <jglisse@redhat.com>
 2012-10-10 00:16:28 +02:00