... and (a >= c) || (b >= c) as max(a, b) >= c.
Similar to commit 97e6c1b9.
total instructions in shared programs: 6182276 -> 6182180 (-0.00%)
instructions in affected programs: 6400 -> 6304 (-1.50%)
helped: 68
HURT: 4
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>
No changes, but does prevent some regressions in the next commit.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>
Helps the same set of programs as the previous commit.
instructions in affected programs: 4490 -> 4346 (-3.21%)
helped: 8
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>
Four shaders in Unreal 4's Sun Temple are helped, and gain SIMD16
because we avoid an integer multiplication.
instructions in affected programs: 2353 -> 2245 (-4.59%)
helped: 4
GAINED: 4
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>
This enables EGL_KHR_fence_sync and EGL_KHR_wait_sync.
Below is the difference in piglit results, before and after this patch.
No regressions and several tests improve from 'skip' to 'pass'. Out of
EGL_KHR_fence_sync tests, two of the multithreaded tests skip; all other
tests pass.
cmdline: piglit run -p gbm -t sync tests/quick.py
mesa: master@1ac7db0
piglit: 4069bec
hw: Ivybridge
| before after
------+-------------
pass | 32 46
fail | 0 0
crash | 0 0
skip | 35 21
total | 67 67
v2:
- Set fence->signalled = true in brw_fence_has_completed() too.
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
I'm about to implement DRI2_Fenc in intel_syncobj.c. To prevent
madness, we need to prefix functions for GL_ARB_sync with 'gl' and
functions for DRI2_Fence with 'dri'. Otherwise, the file will become
a jumble of similiarly named functions.
For example:
old-name: intel_client_wait_sync()
new-name: intel_gl_client_wait_sync()
soon-to-come: intel_dri_client_wait_sync()
I wrote this renaming commit separately from the commit that implements
DRI2_Fence because I wanted the latter diff to be reviewable.
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Don't pass NULL to drm_intel_bo_unreference(). It doesn't like that.
Bug found by code inspection.
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Don't pass NULL to drm_intel_bo_unreference(). It doesn't like that.
Bug found by code inspection.
Reviewed-by: Daniel Stone <daniels@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This should make it more obvious in bug reports while also removing
any sort of guesswork for developers.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Signed-off-by: Martin Peres <martin.peres@linux.intel.com>
Fix build error introduced with commit 1c5a57a "glapi/es3.1: Add support
for GLES versions > 3.0" with Python < 2.7.
File "src/mapi/glapi/gen/gl_genexec.py", line 230, in <module>
printer.Print(api)
File "src/mapi/glapi/gen/gl_XML.py", line 120, in Print
self.printBody(api)
File "src/mapi/glapi/gen/gl_genexec.py", line 187, in printBody
condition_parts.append('(ctx->API == API_OPENGLES2 && ctx->Version >= {})'.format(int(f.api_map['es2'] * 10)))
ValueError: zero length field name in format
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Having the wrong inferred type prevents a number of optimizations,
including constant propagation (since float immediates work differently
than integer immediates).
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Fix Clang return-type error introduced with commit
96f164f6f0 "gallium: make
pipe_context::begin_query return a boolean".
CC r600_query.lo
r600_query.c:443:3: error: non-void function 'r600_begin_query' should return a value [-Wreturn-type]
return;
^
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
This especially helps with NIR because we currently emit MOVs at the top
of the shader to copy from various ATTR registers to a giant VGRF array
of all inputs. (This could potentially be done better, but since
there's only ever one write to each register, it should be trivial to
copy propagate away...)
With NIR - only vertex shaders:
total instructions in shared programs: 3129373 -> 2889581 (-7.66%)
instructions in affected programs: 3119717 -> 2879925 (-7.69%)
helped: 20833
Without NIR - only vertex shaders:
total instructions in shared programs: 2745901 -> 2724483 (-0.78%)
instructions in affected programs: 693426 -> 672008 (-3.09%)
helped: 3516
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
The effective_width field was an ill-concieved hack to get around issues in
the LOAD_PAYLOAD instruction. Now that the LOAD_PAYLOAD instruction is far
more sane, this field can die.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
The newly reworked instruction is far more straightforward than the
original. Before, the LOAD_PAYLOAD instruction was lowered by a the
complicated and broken-by-design pile of heuristics to try and guess
force_writemask_all, exec_size, and a number of other factors on the
sources.
Instead, we use the header_size on the instruction to denote which sources
are "header sources". Header sources are required to be a single physical
hardware register that is copied verbatim. The registers that follow are
considered the actual payload registers and have a width that correspond's
to the LOAD_PAYLOAD's exec_size and are treated as being per-channel. This
gives us a fairly straightforward lowering:
1) All header sources are copied directly using force_writemask_all and,
since they are guaranteed to be a single register, there are no
force_sechalf issues.
2) All non-header sources are copied using the exact same force_sechalf
and force_writemask_all modifiers as the LOAD_PAYLOAD operation itself.
3) In order to accommodate older gens that need interleaved colors,
lower_load_payload detects when the destination is a COMPR4 register
and automatically interleaves the non-header sources. The
lower_load_payload pass does the right thing here regardless of whether
or not the hardware actually supports COMPR4.
This patch commit itself is made up of a bunch of smaller changes squashed
together. Individual change descriptions follow:
i965/fs: Rework fs_visitor::LOAD_PAYLOAD
We rework LOAD_PAYLOAD to verify that all of the sources that count as
headers are, indeed, exactly one register and that all of the non-header
sources match the destination width. We then take the exec_size for
LOAD_PAYLOAD directly from the destination width.
i965/fs: Make destinations of load_payload have the appropreate width
i965/fs: Rework fs_visitor::lower_load_payload
v2: Don't allow the saturate flag on LOAD_PAYLOAD instructions
i965/fs_cse: Support the new-style LOAD_PAYLOAD
i965/fs_inst::is_copy_payload: Support the new-style LOAD_PAYLOAD
i965/fs: Simplify setup_color_payload
Previously, setup_color_payload was a a big helper function that did a
lot of gen-specific special casing for setting up the color sources of
the LOAD_PAYLOAD instruction. Now that lower_load_payload is much more
sane, most of that complexity isn't needed anymore. Instead, we can do
a simple fixup pass for color clamps and then just stash sources
directly in the LOAD_PAYLOAD. We can trust lower_load_payload to do the
right thing with respect to COMPR4.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This commit adds a new is_copy_payload helper to fs_inst that takes the
place of the similarly named functions in cse and register coalesce. The
two is_copy_payload functions in CSE and register coalesce were subtly
different and potentially subtly broken. The new version unifies the two
and should be more correct.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
v2: Get rid of the block parameter and make src a const reference
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Previously, we had a special case for uniforms and immediates and then a
bunch of asserts for various other pessimal things. This commit changes it
so that it explicitly does something on each register file. Some of them
are disallowed and others are treated properly.
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Was off-by-one. llvm says inserting an element with an index higher than the
number of elements yields undefined results. Previously such inserts were
ignored but as of llvm revision 235854 the vector gets replaced with undef,
causing failures.
This fixes piglit gl-3.2-layered-rendering-gl-layer, as mentioned in
https://llvm.org/bugs/show_bug.cgi?id=23424.
Reviewed-by: Brian Paul <brianp@vmware.com>
Cc: mesa-stable@lists.freedesktop.org
This was missing from my patchset to support the query-related entry
points of Direct State Access.
Reported-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Martin Peres <martin.peres@linux.intel.com>
Increases pass rate of ES31-CTS.*program_interface_query* tests
when run with MESA_EXTENSION_OVERRIDE='GL_ARB_compute_shader'. Many
of the negative tests that happen to use compute stage in queries
start passing.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>
Refactoring done on active attribute queries did not take in to
account special built-in inputs for the vertex stage. This commit
sets them referenced by vertex stage so that they get enumerated
properly.
Fixes Piglit test 'get-active-attrib-returns-all-inputs' failure.
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90243
Acked-by: Jose Fonseca <jfonseca@vmware.com>
Tested-by: Dieter Nützel <Dieter@nuetzel-hh.de>
Reviewed-By: Martin Peres <martin.peres@linux.intel.com>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Martin Peres <martin.peres@free.fr>
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Martin Peres <martin.peres@free.fr>
Other drivers which want to enable this extension must expose groups of
GPU hardware performance counters.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>
This patch defines "Driver statistics" and "MP counters" groups, but
only the latter will be exposed through GL_AMD_performance_monitor.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>
This is based on the original patch of Christoph Bumiller.
v2 (Samuel Pitoiset):
- improve Gallium interface for this extension
- rewrite some parts of the original code
- fix compilation errors and piglit tests
v3:
- only enable this extension when the underlying driver expose GPU counters
- get rid of the ring buffer of queries
v4:
- add a debug message when the maximum number of counters has been
reached
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>
GL_AMD_performance_monitor must return an error when a monitoring
session cannot be started.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>
This allows queries to return different numeric types.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>
This will be used by GL_AMD_performance_monitor.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Martin Peres <martin.peres@free.fr>
Matt Turner
Chad Versace
Martin Peres
Vinson Lee
Ilia Mirkin
Ian Romanick
Kenneth Graunke
Jason Ekstrand
Francisco Jerez
Jeremy Huddleston Sequoia
Roland Scheidegger
Chia-I Wu
Tapani Pälli
Chris Forbes
Brian Paul
Samuel Pitoiset
Christoph Bumiller