| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
| |
Use _mesa_program_resource_location_index to fetch index.
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Ilia Mirkin <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
| |
Use program_resource_location to fetch location.
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
| |
Use program_resource_location to fetch location.
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
| |
Instead of iterating IR, retrieve required information through
the new program resource functions.
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Ilia Mirkin <[email protected]>
|
|
|
|
|
|
|
| |
(and mark it as DONE in docs/GL3.txt + 10.6.0 relnotes)
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch adds required helper functions to shaderapi.h and
the actual implementation.
The property query functionality can be tested with tests for
following functions that are refactored by later patches:
GetActiveAtomicCounterBufferiv
GetActiveUniformBlockiv
GetActiveUniformsiv
v2: code cleanup (Ilia Mirkin)
add bufSize < 0 check and error out
fix is_resource_referenced to return bool
check for propCount and bufSize, fixes in buffer_prop
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch adds required helper functions to shaderapi.h and
the actual implementation.
The added functionality can be tested by tests for following
functions that are refactored by later patches:
GetFragDataIndex
v2: return -1 if output not referenced by fragment stage
(Ilia Mirkin)
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch adds required helper functions to shaderapi.h and
the actual implementation.
corresponding Piglit test:
arb_program_interface_query-resource-location
The added functionality can be tested by tests for following
functions that are refactored by later patches:
GetAttribLocation
GetUniformLocation
GetFragDataLocation
v2: code cleanup, changes to array element
syntax checking (Ilia Mirkin)
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch adds required helper functions to shaderapi.h and
the actual implementation.
Name generation copied from '_mesa_get_uniform_name' which can
be removed later by refactoring functions to use resource list.
The added functionality can be tested by tests for following
functions that are refactored by later patches:
GetActiveUniformName
GetActiveUniformBlockName
v2: no index for geometry shader inputs (Ilia Mirkin)
add bufSize < 0 check and error out
validate enum
corresponding Piglit test:
arb_program_interface_query-getprogramresourcename
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch adds required helper functions to shaderapi.h and
the actual implementation.
v2: code cleanup (Ilia Mirkin)
corresponding Piglit test:
arb_program_interface_query-getprogramresourceindex
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch adds required helper functions to shaderapi.h and
the actual implementation.
v2: code cleanup (Ilia Mirkin)
fix array size fo xfb varyings
validate programInterface and throw error
v3: put GL_MAX_NUM_COMPATIBLE_SUBROUTINES where
it belongs
corresponding Piglit test:
arb_program_interface_query-getprograminterfaceiv
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Patch adds ProgramResourceList to gl_shader_program structure.
List contains references to active program resources and is
constructed during linking phase.
This list will be used by follow-up patches to implement hooks
for GL_ARB_program_interface_query. It can be also used to
implement any of the older shader program query APIs.
v2: code cleanups + note for SSBO and subroutines (Ilia Mirkin)
v3: code cleanups + assert(MESA_SHADER_STAGES < 8) (Martin Peres)
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
v2: update dispatch_sanity test (Jason Ekstrand)
+ small code cleanups
v3: xml and Makefile fixes (Ilia Mirkin, Matt Turner)
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously linker did not take in to account case where one would
have only gs and fs (with SSO), patch adds the case by refactoring
code around assign_varying_locations. This makes sure locations for
gs get populated correctly.
This was found with some of the SSO subtests of Martin's upcoming
GetProgramInterfaceiv Piglit test which passes with the patch, no
Piglit regressions.
v2: code cleanups (Martin Peres)
Signed-off-by: Tapani Pälli <[email protected]>
Reviewed-by: Martin Peres <[email protected]>
|
|
|
|
|
|
|
|
|
| |
Fixes assert triggered by
ext_transform_feedback-intervening-read output use_gs
piglit test.
Signed-off-by: Glenn Kennard <[email protected]>
Signed-off-by: Dave Airlie <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Shader-db results:
GM45 NIR:
total instructions in shared programs: 4082044 -> 4081919 (-0.00%)
instructions in affected programs: 27609 -> 27484 (-0.45%)
helped: 44
Iron Lake NIR:
total instructions in shared programs: 5678776 -> 5678646 (-0.00%)
instructions in affected programs: 27406 -> 27276 (-0.47%)
helped: 45
Sandy Bridge NIR:
total instructions in shared programs: 7329995 -> 7329096 (-0.01%)
instructions in affected programs: 142035 -> 141136 (-0.63%)
helped: 406
HURT: 19
Ivy Bridge NIR:
total instructions in shared programs: 6769314 -> 6768359 (-0.01%)
instructions in affected programs: 140820 -> 139865 (-0.68%)
helped: 423
HURT: 2
Haswell NIR:
total instructions in shared programs: 6183693 -> 6183298 (-0.01%)
instructions in affected programs: 96538 -> 96143 (-0.41%)
helped: 303
HURT: 4
Broadwell NIR:
total instructions in shared programs: 7501711 -> 7498170 (-0.05%)
instructions in affected programs: 266403 -> 262862 (-1.33%)
helped: 705
HURT: 5
GAINED: 4
v2: Rebase on top of Connor's fix.
v3: Convert the if-test for num_inputs == 2 to an assertion. Suggested
by Jason after some comments / questions by Ilia.
Signed-off-by: Ian Romanick <[email protected]>
Reviewed-by: Jordan Justen <[email protected]> [v1]
Reviewed-by: Jason Ekstrand <[email protected]>
Cc: Connor Abbott <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously GLX_EXT_create_context_es2_profile was marked as "direct
only" so that it would not depend on server support. Since the
extension required functions that are part of
GLX_ARB_create_context_profile, support for the EXT was disabled if the
ARB was not supported.
This was complete rubbish. If the server supported the ARB but not the
EXT, sending a request with GLX_CONTEXT_ES2_PROFILE_BIT_EXT would result
in GLXBadProfileARB.
Instead of the misguided hack, make GLX_EXT_create_context_es2_profile
properly depend on server support by not marking it as "direct only."
Signed-off-by: Ian Romanick <[email protected]>
Acked-by: José Fonseca <[email protected]>
Reviewed-by: Chad Versace <[email protected]>
Cc: Emil Velikov <[email protected]>
|
|
|
|
|
|
| |
We could potentially support the right combination of 8888 to 565, but the
important thing for now is to not mix up our orderings of 8888. Fixes
fbo-copyteximage regressions.
|
|
|
|
|
| |
Fixes regressions in fbo-generatemipmap-formats on depth/stencil (which
does blits to work around baselevel/lastlevel).
|
|
|
|
|
| |
We don't know who else has written to it, so we'd better update it every
time. This makes the gears spin in X again.
|
| |
|
|
|
|
|
|
| |
As with previous patch.
Reviewed-by: José Fonseca <[email protected]>
|
|
|
|
|
|
|
| |
Now, if we set MESA_LOG_FILE and MESA_GLSL=dump, all the shader info
will get logged to the named file instead of stderr.
Reviewed-by: José Fonseca <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
_mesa_log() simply writes log information to stderr or MESA_LOG_FILE.
_mesa_get_log_file() returns the file handle to use for logging.
This will be used for shader dumping/logging instead of always printing
to stderr.
Reviewed-by: José Fonseca <[email protected]>
|
|
|
|
|
|
| |
So we can see the label associated with subroutines.
Reviewed-by: José Fonseca <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
dispatch.
- Use GetModuleHandle instead of LoadLibrary to avoid incrementing the
opengl32.dll reference count (otherwise the opengl32.dll will linger
in memory forever.)
- Ensure we use our fake wglCreateContext/wglDeleteContext when using
Mesa as a drop-in replacement for opengl32.dll
Untested. Just noticed by accident.
Reviewed-by: Brian Paul <[email protected]>
|
|
|
|
|
|
| |
As pointed out by Shervin Sharifi.
Reviewed-by: Brian Paul <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When a vec has more elements than row components in a matrix, the
code could end up failing an assert inside assign_to_matrix_column().
This patch makes sure that when there is still room in the matrix for
more elements (but in other columns of the matrix), the data is actually
assigned.
This patch fixes the following dEQP test:
dEQP-GLES3.functional.shaders.conversions.matrix_combine.float_bvec4_ivec2_bool_to_mat4x2_vertex
dEQP-GLES3.functional.shaders.conversions.matrix_combine.float_bvec4_ivec2_bool_to_mat4x2_fragment
Signed-off-by: Samuel Iglesias Gonsalvez <[email protected]>
Reviewed-by: Ben Widawsky <[email protected]>
|
|
|
|
|
|
|
|
| |
Signed-off-by: Ian Romanick <[email protected]>
Reviewed-by: Jordan Justen <[email protected]>
Reviewed-by: Jason Ekstrand <[email protected]>
Reviewed-by: Connor Abbott <[email protected]>
Cc: "10.5" <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
nir/nir.h: In function 'nir_validate_shader':
nir/nir.h:1567:56: warning: unused parameter 'shader' [-Wunused-parameter]
static inline void nir_validate_shader(nir_shader *shader) { }
^
nir/nir_opt_cse.c: In function 'src_is_ssa':
nir/nir_opt_cse.c:165:32: warning: unused parameter 'data' [-Wunused-parameter]
src_is_ssa(nir_src *src, void *data)
^
nir/nir_opt_cse.c: In function 'dest_is_ssa':
nir/nir_opt_cse.c:171:35: warning: unused parameter 'data' [-Wunused-parameter]
dest_is_ssa(nir_dest *dest, void *data)
^
Signed-off-by: Ian Romanick <[email protected]>
Reviewed-by: Jordan Justen <[email protected]>
Reviewed-by: Connor Abbott <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
We weren't comparing the right number of components when checking
swizzles. Use nir_ssa_alu_instr_num_src_components() to do the right
thing.
No piglit regressions, and no fixes either.
Reviewed-by: Ian Romanick <[email protected]>
Reviewed-by: Jason Ekstrand <[email protected]>
Signed-off-by: Connor Abbott <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Certain platforms support the ability to sample from a texture, and write it out
to the file RT - thus saving a costly send instructions (note that this is a
potnential win if one wanted to backport to a tag that didn't have the patch
from Topi which removed excess MOVs from LOAD_PAYLOAD - 97caf5fa04dbd2),
v2: Modify the algorithm. Instead of iterating in reverse through blocks and
insts, since the last block/inst is the only thing which can benefit. Rebased
on top of Ken's patching modifying is_last_send
v3: Rebased over almost 2 months, and Incorporated feedback from Matt:
Some comment typo fixes and rewordings.
Whitespace
Move the optimization pass outside of the optimize loop
v4: Some cosmetic changes requested from Ken. These changes ensured that the
optimization function always returned true when an optimization occurred, and
false when one did not. This behavior did not exist with the original patch. As
a result, having the separate helper function which Matt did not like no longer
made sense, and so now I believe everyone should be happy.
Benchmark (n=20) %diff
*OglBatch5 -1.4
*OglBatch7 -1.79
OglFillTexMulti 5.57
OglFillTexSingle 1.16
OglShMapPcf 0.05
OglTexFilterAniso 3.01
OglTexFilterTri 1.94
No piglit regressions:
(http://otc-gfxtest-01.jf.intel.com:8080/view/dev/job/bwidawsk/112/)
[*] I believe my measurements are incorrect for Batch5-7. If I add this new
optimization, but never emit the new instruction I see similar results.
v5: Remove declaration of combine_tex_header since v4 dropped that function
(Ben)
Remove check for impossible case of an empty block (Matt)
Set dest earlier to avoid extra special-casing in generate_tex (Matt)
Signed-off-by: Ben Widawsky <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Based originally on a patch from Ken in May 2014 of the same title. Things
changed enough that I didn't feel comfortable leaving his authorship.
v2: Replace fp->UsesKill with wm_prog_data->uses_kill. Since Ken took the time
to also explain the difference to me, here is his explanation for posterity:
"fp->UsesKill indicates that a ARB_fragment_program shader uses the KIL
instruction, or that a GLSL shader uses the "discard" insntruction
(which are analogous).
On Gen4-5, we sometimes have to simulate OpenGL's "Alpha Test" feature
by emitting shader code that implicitly does a "discard" instruction.
In the key setup, we do:
/* key->alpha_test_func means simulating alpha testing via discards,
* so the shader definitely kills pixels.
*/
prog_data.uses_kill = fp->program.UsesKill || key->alpha_test_func;
Even though the shader may not technically contain a "discard", we need
to act as if it does.
I've also been trying to move the i965 state setup code to use
brw_wm_prog_key for everything, rather than poking at core Mesa's
gl_program/gl_fragment_program/gl_shader/gl_shader_program structures.
--Ken"
Signed-off-by: Ben Widawsky <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When an instruction has a side effect, it impacts the available options when
reordering an instruction. As the EOT flag is an implied write to the render
target in the FS, it can be considered a side effect.
This patch shouldn't actually have any impact on the current code since the EOT
flag implies that the opcode is already one with side effects,
FS_OPCODE_FB_WRITE. The next patch however will introduce an optimization
whereby the EOT flag can occur with an opcode SHADER_OPCODE_TEX, and as that
instruction will perform the same implied write to the render target, it cannot
be reordered.
v2: Remove extra whitespace (Matt)
Signed-off-by: Ben Widawsky <[email protected]>
Reviewed-by: Kenneth Graunke <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
The patch was verified with Microsoft Visual studio 2013
redistributable package(RTM version number: 18.0.21005.1)
Next MSVC versions intends to support __func__.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
The patch was verified with Microsoft Visual studio 2013
redistributable package(RTM version number: 18.0.21005.1)
Next MSVC versions intends to support __func__.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
The patch was verified with Microsoft Visual studio 2013
redistributable package(RTM version number: 18.0.21005.1)
Next MSVC versions intends to support __func__.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
The patch was verified with Microsoft Visual studio 2013
redistributable package(RTM version number: 18.0.21005.1)
Next MSVC versions intends to support __func__.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Reviewed-by: Anuj Phogat <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
Consistently just use C99's __func__ everywhere.
The patch was verified with Microsoft Visual studio 2013
redistributable package(RTM version number: 18.0.21005.1)
Next MSVC versions intends to support __func__.
No functional changes.
Acked-by: Matt Turner <[email protected]>
Reviewed-by: Anuj Phogat <[email protected]>
Signed-off-by: Marius Predut <[email protected]>
|
|
|
|
|
|
|
|
|
| |
a * b is equivalent to -a * -b, and the previous code was failing at
that.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89961
Reviewed-by: Anuj Phogat <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit e93566a15c61c33faa changed the message header code needed to
make Skylake use SIMD4x2 so that it uses a register with width 4
instead of 8 as the source register in the send message. However it
also changed the width for the dest in the MOV instruction which is
used to initialise the header register with the values from g0. The
width of the destination is used to determine the exec size in
brw_set_dest so this would end up making the MOV have an exec size of
4. I think this would end up leaving the top half of the register
uninitialised. The top half of the header has meaningful values so
this probably isn't a good idea.
This patch just casts the dest register for the MOV instruction back
to a vec8 to fix it. It doesn't cause any changes to a Piglit run.
Reviewed-by: Ben Widawsky <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Commit b616164 added an optimization of b2f generation of a comparison.
It also included an extra optimization of one of the comparison values
is a constant of zero. The trick was that some value was known to be
zero, so that value could be used in the SEL instruction instead of
potentially loading 0.0 into a register.
This change switched the order of the arguments to the SEL, and, for
some unknown reason, I thought that the predicate should therefore
only be inverted for the == case. Clearly, it should always be
inverted.
Fixes piglit fs-notEqual-of-expression.shader_test and
fs-equal-of-expression.shader_test.
v2: Don't do the "register already has zero" optimization for the '== 0'
case. In that case, the register does not have zero when we want to
produce a zero result.
Signed-off-by: Ian Romanick <[email protected]>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=89722
Reviewed-by: Kenneth Graunke <[email protected]> [v1]
Tested-by: Lu Hua <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Rather than emitting one kill instruction per component of KILL_IF's src
reg, we now or the components of the src register together and use the
result as a condition for just one kill instruction.
shader-db stats (bonaire):
979 shaders
Totals:
SGPRS: 34872 -> 34848 (-0.07 %)
VGPRS: 20696 -> 20676 (-0.10 %)
Code Size: 749032 -> 748452 (-0.08 %) bytes
LDS: 11 -> 11 (0.00 %) blocks
Scratch: 12288 -> 12288 (0.00 %) bytes per wave
Totals from affected shaders:
SGPRS: 1184 -> 1160 (-2.03 %)
VGPRS: 600 -> 580 (-3.33 %)
Code Size: 13200 -> 12620 (-4.39 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
Increases:
SGPRS: 2 (0.00 %)
VGPRS: 0 (0.00 %)
Code Size: 0 (0.00 %)
LDS: 0 (0.00 %)
Scratch: 0 (0.00 %)
Decreases:
SGPRS: 5 (0.01 %)
VGPRS: 5 (0.01 %)
Code Size: 25 (0.03 %)
LDS: 0 (0.00 %)
Scratch: 0 (0.00 %)
*** BY PERCENTAGE ***
Max Increase:
SGPRS: 32 -> 40 (25.00 %)
VGPRS: 0 -> 0 (0.00 %)
Code Size: 0 -> 0 (0.00 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
Max Decrease:
SGPRS: 32 -> 24 (-25.00 %)
VGPRS: 16 -> 12 (-25.00 %)
Code Size: 116 -> 96 (-17.24 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
*** BY UNIT ***
Max Increase:
SGPRS: 64 -> 72 (12.50 %)
VGPRS: 0 -> 0 (0.00 %)
Code Size: 0 -> 0 (0.00 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
Max Decrease:
SGPRS: 32 -> 24 (-25.00 %)
VGPRS: 16 -> 12 (-25.00 %)
Code Size: 424 -> 356 (-16.04 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This should improve code quality in general and will help with some
future changes to how we emit kill instructions.
shader-db shows a few regressions, but these don't seem to be the result
of deficiencies in instcombine. They're mostly caused by the scheduler
making different decisions than before.
shader-db stats (bonaire):
979 shaders
Totals:
SGPRS: 35056 -> 34872 (-0.52 %)
VGPRS: 20624 -> 20696 (0.35 %)
Code Size: 764372 -> 749032 (-2.01 %) bytes
LDS: 11 -> 11 (0.00 %) blocks
Scratch: 12288 -> 12288 (0.00 %) bytes per wave
Totals from affected shaders:
SGPRS: 13264 -> 13072 (-1.45 %)
VGPRS: 8248 -> 8316 (0.82 %)
Code Size: 486320 -> 470992 (-3.15 %) bytes
LDS: 11 -> 11 (0.00 %) blocks
Scratch: 11264 -> 11264 (0.00 %) bytes per wave
Increases:
SGPRS: 6 (0.01 %)
VGPRS: 20 (0.02 %)
Code Size: 14 (0.01 %)
LDS: 0 (0.00 %)
Scratch: 0 (0.00 %)
Decreases:
SGPRS: 32 (0.03 %)
VGPRS: 8 (0.01 %)
Code Size: 244 (0.25 %)
LDS: 0 (0.00 %)
Scratch: 0 (0.00 %)
*** BY PERCENTAGE ***
Max Increase:
SGPRS: 32 -> 48 (50.00 %)
VGPRS: 12 -> 20 (66.67 %)
Code Size: 216 -> 224 (3.70 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
Max Decrease:
SGPRS: 40 -> 32 (-20.00 %)
VGPRS: 16 -> 12 (-25.00 %)
Code Size: 368 -> 280 (-23.91 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
*** BY UNIT ***
Max Increase:
SGPRS: 32 -> 48 (50.00 %)
VGPRS: 28 -> 36 (28.57 %)
Code Size: 39320 -> 40132 (2.07 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
Max Decrease:
SGPRS: 72 -> 64 (-11.11 %)
VGPRS: 48 -> 40 (-16.67 %)
Code Size: 6272 -> 5852 (-6.70 %) bytes
LDS: 0 -> 0 (0.00 %) blocks
Scratch: 0 -> 0 (0.00 %) bytes per wave
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
| |
This makes it easier to parse.
Reviewed-by: Marek Olšák <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
new_prim was declared as a stack variable within a nested scope; we
tried to retain a pointer to that data beyond the scope, which is bogus.
GCC with -O1 eliminated most of the code that set new_prim's fields.
Move the declaration to fix the bug.
v2: Also fix new_ib (thanks to Matt Turner and Ben Widawsky).
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=81025
Signed-off-by: Kenneth Graunke <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
Reviewed-by: Ben Widawsky <[email protected]>
Cc: [email protected]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
I finally managed to dig up some information on our mysterious GPU hangs.
A wiki page from the Crestline validation team mentions that they found
a GPU hang in "Serious Sam 2" (on Windows) with remarkably similar
conditions to the ones we've seen in Google Chrome and glmark2.
Apparently, if WM_STATE has "PS Use Source Depth" enabled, CC_STATE has
most depth state disabled, and you issue a CONSTANT_BUFFER command and
immediately draw, the depth interpolator makes a small mistake that
leads to hangs.
Most of the traces I looked at contained a CONSTANT_BUFFER packet
immediately followed by 3DPRIMITIVE, or at least very few packets.
It appears they also have "PS Use Source Depth" enabled - either at the
hang, or a little before it. So I think this is our bug.
The workaround is to emit a non-pipelined state packet after issuing a
CONSTANT_BUFFER packet. This is really similar to the workaround I
developed in commit c4fd0c9052dd391d6f2e9bb8e6da209dfc7ef35b.
v2: Fix word-wrapping issues.
Signed-off-by: Kenneth Graunke <[email protected]>
Reviewed-by: Francisco Jerez <[email protected]>
Reviewed-by: Matt Turner <[email protected]>
|