---
src/mesa/drivers/dri/radeon/radeon_screen.h | 12
src/util/xmlpool/ca.po | 17 -
src/util/xmlpool/de.po | 17 -
src/util/xmlpool/es.po | 17 -
src/util/xmlpool/fr.po
---
src/mesa/drivers/dri/radeon/radeon_screen.h | 13 +
src/util/xmlpool/ca.po | 20
src/util/xmlpool/de.po | 20
src/util/xmlpool/es.po | 20
---
src/mesa/drivers/dri/radeon/radeon_screen.c | 5 +
src/util/xmlpool/ca.po | 4
src/util/xmlpool/de.po | 4
src/util/xmlpool/es.po | 4
src/util/xmlpool/fr.po | 4
src/util/xmlpool/nl.po
---
src/mesa/drivers/dri/radeon/radeon_screen.h | 10 ++
src/util/xmlpool/ca.po | 12
src/util/xmlpool/de.po | 12
src/util/xmlpool/es.po | 12
src/util/xmlpool/fr.po
---
src/mesa/drivers/dri/radeon/radeon_screen.c | 5 +
src/util/xmlpool/ca.po | 4
src/util/xmlpool/de.po | 4
src/util/xmlpool/es.po | 4
src/util/xmlpool/fr.po | 4
src/util/xmlpool/nl.po
This seems to have only been used by DRI1 drivers which were
removed with e4344161bde2.
---
src/util/xmlpool/t_options.h | 5 -
1 file changed, 5 deletions(-)
diff --git a/src/util/xmlpool/t_options.h b/src/util/xmlpool/t_options.h
index 3e5993caf78..f0dc78dd595 100644
---
This seems to have only been used by DRI1 drivers which were
removed with e4344161bde2.
---
src/util/xmlpool/ca.po | 16
src/util/xmlpool/de.po | 16
src/util/xmlpool/es.po | 16
src/util/xmlpool/fr.po | 16
This seems to have only been used by DRI1 drivers which were
removed with e4344161bde2.
---
src/util/xmlpool/ca.po | 4
src/util/xmlpool/de.po | 4
src/util/xmlpool/es.po | 4
src/util/xmlpool/fr.po | 4
src/util/xmlpool/nl.po | 4
---
src/mesa/drivers/dri/radeon/radeon_screen.c | 5 +
src/util/xmlpool/ca.po | 4
src/util/xmlpool/de.po | 4
src/util/xmlpool/es.po | 4
src/util/xmlpool/fr.po | 4
src/util/xmlpool/nl.po
---
src/mesa/drivers/dri/radeon/radeon_screen.h | 12
src/util/xmlpool/ca.po | 16
src/util/xmlpool/de.po | 16
src/util/xmlpool/es.po | 16
src/util/xmlpool/fr.po
---
src/mesa/drivers/dri/radeon/radeon_screen.c | 6 ++
src/util/xmlpool/ca.po | 6 --
src/util/xmlpool/de.po | 6 --
src/util/xmlpool/es.po | 6 --
src/util/xmlpool/fr.po | 5 -
---
src/util/xmlpool/ca.po | 7 ---
src/util/xmlpool/de.po | 7 ---
src/util/xmlpool/es.po | 7 ---
src/util/xmlpool/fr.po | 6 --
src/util/xmlpool/nl.po | 7 ---
src/util/xmlpool/sv.po | 5 -
6 files changed, 39 deletions(-)
diff --git a/src/util/xmlpool/ca.po
Use enviroment var overrides in legacy drivers instead.
---
src/mesa/drivers/dri/r200/r200_context.c | 4 +---
src/mesa/drivers/dri/radeon/radeon_context.c | 5 ++---
src/mesa/drivers/dri/radeon/radeon_screen.c | 2 --
src/util/xmlpool/ca.po | 23
---
src/mesa/drivers/dri/radeon/radeon_screen.h | 10 ++
src/util/xmlpool/ca.po | 12
src/util/xmlpool/de.po | 12
src/util/xmlpool/es.po | 12
src/util/xmlpool/fr.po
This seems to have only been used by DRI1 drivers which were
removed with e4344161bde2.
---
src/util/xmlpool/ca.po | 4
src/util/xmlpool/de.po | 4
src/util/xmlpool/es.po | 4
src/util/xmlpool/fr.po | 4
src/util/xmlpool/nl.po | 4
Forcing software fallbacks for i965 hasn't been an option since
5e3c093ff866.
---
src/mesa/drivers/dri/i965/brw_context.h | 1 -
1 file changed, 1 deletion(-)
diff --git a/src/mesa/drivers/dri/i965/brw_context.h
b/src/mesa/drivers/dri/i965/brw_context.h
index 72be8f2a4d0..c32def7c3d7 100644
---
These seems to have only been used by DRI1 drivers which were
removed with e4344161bde2.
---
src/util/xmlpool/ca.po | 8
src/util/xmlpool/de.po | 8
src/util/xmlpool/es.po | 8
src/util/xmlpool/fr.po | 8
src/util/xmlpool/nl.po
This driver is in maintenance mode so lets remove this hidden
unsafe option.
---
src/mesa/drivers/dri/i915/i915_vtbl.c | 8
src/mesa/drivers/dri/i915/intel_context.c | 2 --
src/mesa/drivers/dri/i915/intel_context.h | 2 --
src/mesa/drivers/dri/i915/intel_screen.c | 4
4 files
This series removes some unused options, replaces some legacy
debug options with env vars instead and moves a bunch of old
radeon dri driver options directly into the drivers.
I haven't bothered copying all the radeon dri translations
as I didn't want to waste my time as it seems unlikely anyone
This was added as a workaround for Heaven 3.0 but was later removed
by 5ead448719f3 to allow Heaven 4.0 to work correctly.
---
src/gallium/auxiliary/pipe-loader/driinfo_gallium.h | 1 -
src/gallium/include/state_tracker/st_api.h | 1 -
src/gallium/state_trackers/dri/dri_screen.c
Add enviroment var overrides to legacy drivers instead.
---
src/mesa/drivers/dri/i915/intel_context.c| 2 +-
src/mesa/drivers/dri/i915/intel_screen.c | 1 -
src/mesa/drivers/dri/i965/intel_screen.c | 1 -
src/mesa/drivers/dri/r200/r200_context.c | 2 +-
On 09/08/18 05:05, Jason Ekstrand wrote:
Commit 4434591bf56a6b0 caused substantially more URB messages in
geometry and tessellation shaders. Before we can really enable this
sort of optimization, We either need some way of combining them back
together into vectors or we need to do cross-stage
patches 1-2:
Reviewed-by: Timothy Arceri
On 11/08/18 04:32, Marek Olšák wrote:
From: Marek Olšák
---
src/gallium/drivers/radeonsi/si_get.c | 49 ++
src/gallium/drivers/radeonsi/si_pipe.c | 2 +-
src/gallium/drivers/radeonsi/si_pipe.h | 1 -
3 files changed
I really only skimmed over patch 9. But everything seemed ok so for the
series:
Reviewed-by: Timothy Arceri
On 08/08/18 15:42, Marek Olšák wrote:
From: Marek Olšák
---
docs/relnotes/18.3.0.html | 1 +
src/mesa/state_tracker/st_extensions.c | 1 +
2 files changed, 2
6-10:
Reviewed-by: Timothy Arceri
On 09/08/18 23:43, Alejandro Piñeiro wrote:
From: Neil Roberts
Instead of using the copy of shader_info stored in gl_program, it now
uses the one in nir_shader. This is needed for SPIR-V because the
info.tess.tcs_vertices_out is filled in via
Reviewed-by: Timothy Arceri
On 09/08/18 23:43, Alejandro Piñeiro wrote:
Equivalent to the already existing how_declared at GLSL IR. The only
difference is that we are not adding all the declaration_type
available on GLSL, only the one that we will use on the short term. We
would add more mode
Reviewed-by: Timothy Arceri
On 09/08/18 23:43, Alejandro Piñeiro wrote:
So they are not exposed through the introspection API.
It is worth to note that the number of hidden uniforms of GLSL linking
vs SPIR-V linking would be somewhat different due the differen order
of the nir lowerings
Reviewed-by: Timothy Arceri
On 09/08/18 23:43, Alejandro Piñeiro wrote:
info.gs.output_primitive was already being filled. Not sure why this
is not needed on Vulkan, but we found to be needed for
ARB_gl_spirv. Specifically, this is needed to get the following test
passing:
KHR-GL45
1, 2018 at 2:39 AM, Dieter Nützel
wrote:
Am 01.08.2018 05:56, schrieb Timothy Arceri:
On 01/08/18 13:09, Dieter Nützel wrote:
Am 31.07.2018 13:34, schrieb Timothy Arceri:
On 31/07/18 13:50, Dieter Nützel wrote:
Am 30.07.2018 05:24, schrieb Dieter Nützel:
For the series
I thought you were trying to move these type of opts to your new pass?
Or should I be splitting out and resending some of the bcsel opts in my
dirt showdown patch [1]? And trying again to get the other patches from
the series reviewed.
[1]
This stops Coverity reporting a defect and helps make the code less
error-prone.
Cc: Neil Roberts
---
src/compiler/nir/nir_gather_info.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/compiler/nir/nir_gather_info.c
b/src/compiler/nir/nir_gather_info.c
index d0b656ea741..4a030cb6256
On 31/07/18 16:52, Bas Nieuwenhuizen wrote:
On Fri, Jul 27, 2018 at 11:40 AM, Timothy Arceri wrote:
On 27/07/18 19:10, Bas Nieuwenhuizen wrote:
On Fri, Jul 27, 2018 at 7:32 AM, Timothy Arceri
wrote:
Because this was setting image to true we would end up calling
si_load_image_desc
On 01/08/18 13:09, Dieter Nützel wrote:
Am 31.07.2018 13:34, schrieb Timothy Arceri:
On 31/07/18 13:50, Dieter Nützel wrote:
Am 30.07.2018 05:24, schrieb Dieter Nützel:
For the series
Tested-by: Dieter Nützel
with glmark2, glxgears, UH, UV, Blender 2.79 and FreeCAD 0.17
on RX 580
I guess you guys may want to look into Ken's comments on patch 3. But
otherwise the series looks good to me and is:
Reviewed-by: Timothy Arceri
On 01/08/18 05:22, Jason Ekstrand wrote:
---
src/intel/compiler/brw_nir.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src
Reviewed-by: Timothy Arceri
On 31/07/18 21:16, Alejandro Piñeiro wrote:
FWIW, this is the only patch pending to be reviewed on the series (sorry
Timothy, I used wrong patch numbers when I pinged you on IRC).
BR
On 20/07/18 17:08, Alejandro Piñeiro wrote:
---
src/compiler/shader_info.h
.
Are you able to bisect?
Regards,
Dieter
Am 28.07.2018 03:07, schrieb Timothy Arceri:
On 28/07/18 11:06, Timothy Arceri wrote:
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All the spill changes in shader-db are from Dolphin uber shaders
1-4:
Reviewed-by: Timothy Arceri
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
On 24/07/18 23:29, Alejandro Piñeiro wrote:
From: Neil Roberts
v2: ignore names on purpose, for consistency with other places where
we are doing the same (Alejandro)
v3: changes proposed by Timothy Arceri, implemented by Alejandro Piñeiro:
* Remove redundant 'struct
1 << (MESA_SHADER_GEOMETRY + 1)) - 1));
As ((1 << MESA_SHADER_VERTEX) - 1)) == 0
If you use the above simplification this patch is:
Reviewed-by: Timothy Arceri
+ if (last_vert_stage)
+ prog->last_vert_prog = prog->_LinkedShaders[last_vert_stage -
1]->Program;
}
Reviewed-by: Timothy Arceri
On 21/07/18 01:08, Alejandro Piñeiro wrote:
From: Neil Roberts
This just sets the stream on the nir_variable.
---
src/compiler/spirv/vtn_variables.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/src/compiler/spirv/vtn_variables.c
b
On 28/07/18 11:06, Timothy Arceri wrote:
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All the spill changes in shader-db are from Dolphin uber shaders,
despite some small regressions the change is clearly positive.
V2: insert new constant
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All the spill changes in shader-db are from Dolphin uber shaders,
despite some small regressions the change is clearly positive.
V2: insert new constant after any phis in the
eref_path_init(_path, b, NULL);
+ assert(a_path.path[0]->deref_type == nir_deref_type_var);
+ assert(b_path.path[0]->deref_type == nir_deref_type_var);
+
+ nir_deref_compare_result result = nir_compare_deref_paths(_path, _path);
+
+ nir_deref_path_finish(_path);
+ nir_deref_path_finish(_path);
On 27/07/18 02:00, Jason Ekstrand wrote:
This pass looks for array variables where at least one level of the
array is never indirected and splits it into multiple smaller variables.
---
src/compiler/nir/nir.h| 1 +
src/compiler/nir/nir_split_vars.c | 525
On 27/07/18 19:10, Bas Nieuwenhuizen wrote:
On Fri, Jul 27, 2018 at 7:32 AM, Timothy Arceri wrote:
Because this was setting image to true we would end up calling
si_load_image_desc() when we sould be calling
si_load_sampler_desc().
Since the descriptor is part of an image, not a sampler
Because this was setting image to true we would end up calling
si_load_image_desc() when we sould be calling
si_load_sampler_desc().
This fixes an assert() in Deus Ex: MD
---
src/amd/common/ac_nir_to_llvm.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git
Looks like a mistake from when the deref stuff landed.
Fixes: 506a07e4e3a4 ("ac/nir: Add deref support to image intrinsics.")
Cc: Bas Nieuwenhuizen
---
src/amd/common/ac_nir_to_llvm.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/amd/common/ac_nir_to_llvm.c
Any reason there are no shader-db results to go with the commits that
turn on the opts?
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
On 26/07/18 14:04, Timothy Arceri wrote:
On 23/07/18 18:02, Timothy Arceri wrote:
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All the spill changes in shader-db are from Dolphin uber shaders,
despite some small regressions the change
On 23/07/18 18:02, Timothy Arceri wrote:
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All the spill changes in shader-db are from Dolphin uber shaders,
despite some small regressions the change is clearly positive.
shader-db results IVB
It would be good if you could reply to my comment in patch 13 otherwise
1-17 are:
Reviewed-by: Timothy Arceri
As mentioned before you really need the varying vector and array
splitting passes to get the most from these linking opts.
Patch 18 seems ok but you might want to get someone else
On 12/07/18 07:18, Jason Ekstrand wrote:
This pulls the SPIR-V to NIR step out into common code.
---
src/intel/vulkan/anv_pipeline.c | 278 +---
1 file changed, 116 insertions(+), 162 deletions(-)
diff --git a/src/intel/vulkan/anv_pipeline.c
1-8 look fine to me and are:
Reviewed-by: Timothy Arceri
I'll try to finish looking over the rest later tonight or tomorrow.
On 12/07/18 07:18, Jason Ekstrand wrote:
Instead of hashing each stage separately (and TES and TCS together), we
hash the entire pipeline. This means we'll get fewer
On 24/07/18 00:10, Jason Ekstrand wrote:
On Mon, Jul 23, 2018 at 1:08 AM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote:
Ian and I have been looking at these type of things recently. Ian has
started work on a pass to cover this stuff without having to add dozens
of th
Reviewed-by: Timothy Arceri
On 25/07/18 07:55, Kenneth Graunke wrote:
This wasn't updated for the new scalar ISA parameter. It worked anyway
because all the function's callers live in the same file, so it found
the correct function. Tim made this external for the new st prog_to_nir
On 24/07/18 14:24, Marek Olšák wrote:
On Mon, Jul 23, 2018 at 11:33 PM, Timothy Arceri wrote:
On 24/07/18 11:15, Marek Olšák wrote:
On Fri, Jul 20, 2018 at 12:53 AM, Dave Airlie wrote:
On 20 July 2018 at 13:12, Marek Olšák wrote:
From: Marek Olšák
To make dEQP-GLES31
On 24/07/18 11:15, Marek Olšák wrote:
On Fri, Jul 20, 2018 at 12:53 AM, Dave Airlie wrote:
On 20 July 2018 at 13:12, Marek Olšák wrote:
From: Marek Olšák
To make dEQP-GLES31.functional.ssbo.layout.random.all_shared_buffer.23
finish sooner on the older CPUs. (otherwise it gets killed and we
On 23/07/18 22:48, Alejandro Piñeiro wrote:
On 21/07/18 05:15, Timothy Arceri wrote:
On 21/07/18 13:09, Timothy Arceri wrote:
Reviewed-by: Timothy Arceri
Actually I take that back. This introduces a dependency on GL in NIR,
Hmm, but that dependency is already there. nir.h includes GL/gl.h
Ian and I have been looking at these type of things recently. Ian has
started work on a pass to cover this stuff without having to add dozens
of these types of opts.
https://lists.freedesktop.org/archives/mesa-dev/2018-July/200583.html
On 23/07/18 17:36, Jason Ekstrand wrote:
Spotted in a
In order to be sure loop_terminator_list is an accurate
representation of all the jumps in the loop we need to be sure we
didn't encounter any other complex behaviour such as continues,
nested breaks, etc during analysis.
This will be used in the following patch.
---
src/compiler/nir/nir.h
In GLSL IR we cheat with switch statements and simply convert them
into loops with a single iteration. This allowed us to make use of
the existing jump instruction handling provided by the loop handing
code, it also allows dead code to be cleaned up once we have
wrapped the code in a loop.
Now that SSA values can be derefs and they have special rules, we have
to be a bit more careful about our LCSSA phis. In particular, we need
to clean up in case LCSSA ended up creating a phi node for a deref.
This avoids validation issues with some CTS tests with the following
patch, but its
The innermost check was added to stop us from unrolling multiple
loops in a single pass, and to stop outer loops from unrolling.
When we successfully unroll a loop we need to run the analysis
pass again before deciding if we want to go ahead an unroll a
second loop.
However the logic was flawed
This adds support for unrolling the classic
do {
// ...
} while (false)
that is used to wrap multi-line macros. GLSL IR also wraps switch
statements in a loop like this.
shader-db results IVB:
total loops in shared programs: 2515 -> 2512 (-0.12%)
loops in affected programs: 33
This will help later patches with unrolling loops that end with a
break i.e. loops the always exit on their first interation.
---
src/compiler/nir/nir_loop_analyze.c | 14 +++---
1 file changed, 7 insertions(+), 7 deletions(-)
diff --git a/src/compiler/nir/nir_loop_analyze.c
v2:
- make use of nir_block_dominates() in the if-statement opts as suggested
by Connor. This allowed for a reduction in code required but also allowed
me to implemented some extra optimisations.
- implement feedback from Jason on loop unrolling opts.
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All the spill changes in shader-db are from Dolphin uber shaders,
despite some small regressions the change is clearly positive.
shader-db results IVB:
total instructions in shared programs:
shader-db IVB results:
total instructions in shared programs: 9993483 -> 9993472 (-0.00%)
instructions in affected programs: 1300 -> 1289 (-0.85%)
helped: 11
HURT: 0
total cycles in shared programs: 219476091 -> 219476059 (-0.00%)
cycles in affected programs: 7675 -> 7643 (-0.42%)
helped: 10
On 22/07/18 19:50, Jason Ekstrand wrote:
On Tue, Jul 10, 2018 at 11:49 PM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote:
This adds support for unrolling the classic
do {
// ...
} while (false)
that is used to wrap multi-line macros. GLS
On 22/07/18 19:50, Jason Ekstrand wrote:
On Tue, Jul 10, 2018 at 11:49 PM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote:
This adds support for unrolling the classic
do {
// ...
} while (false)
that is used to wrap multi-line macros. GLS
On 22/07/18 19:32, Jason Ekstrand wrote:
On Tue, Jul 10, 2018 at 11:49 PM Timothy Arceri <mailto:tarc...@itsqueeze.com>> wrote:
The innermost check was added to stop us from unrolling multiple
loops in a single pass, and to stop outer loops from unrolling.
Do you mean "m
V2: make use of visit_tree()
Fixes new piglit test:
tests/spec/glsl-1.20/execution/qualifiers/vs-out-conversion-int-to-float-vec4-index.shader_test
---
src/compiler/glsl/ast_function.cpp | 54 ++
1 file changed, 54 insertions(+)
diff --git
Reviewed-by: Timothy Arceri
On 21/07/18 01:08, Alejandro Piñeiro wrote:
---
src/mesa/drivers/dri/i965/brw_context.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/src/mesa/drivers/dri/i965/brw_context.c
b/src/mesa/drivers/dri/i965/brw_context.c
index 968fc1d43d6..136f1325c6e 100644
Reviewed-by: Timothy Arceri
On 21/07/18 01:08, Alejandro Piñeiro wrote:
From: Neil Roberts
---
src/mesa/drivers/dri/i965/brw_link.cpp | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/mesa/drivers/dri/i965/brw_link.cpp
b/src/mesa/drivers/dri/i965/brw_link.cpp
index e7a940e15c2
On 21/07/18 01:08, Alejandro Piñeiro wrote:
From: Neil Roberts
v2: ignore names on purpose, for consistency with other places where
we are doing the same (Alejandro)
---
src/compiler/Makefile.sources | 1 +
src/compiler/glsl/gl_nir_link_xfb.c | 289
On 21/07/18 13:09, Timothy Arceri wrote:
Reviewed-by: Timothy Arceri
Actually I take that back. This introduces a dependency on GL in NIR,
maybe better to put this one in gl_nir_linker.h? If you agree that
change would be:
Reviewed-by: Timothy Arceri
On 21/07/18 01:08, Alejandro
Reviewed-by: Timothy Arceri
On 21/07/18 01:08, Alejandro Piñeiro wrote:
From: Neil Roberts
---
src/compiler/nir_types.cpp | 6 ++
src/compiler/nir_types.h | 2 ++
2 files changed, 8 insertions(+)
diff --git a/src/compiler/nir_types.cpp b/src/compiler/nir_types.cpp
index
Reviewed-by: Timothy Arceri
On 21/07/18 01:08, Alejandro Piñeiro wrote:
For now we are just adding nir lowerings that are needed/mandatory to
get things working. After everything is settled, we would start to add
good-to-have lowerings.
This patch adds the following calls
1-3:
Reviewed-by: Timothy Arceri
On 21/07/18 01:08, Alejandro Piñeiro wrote:
From: Neil Roberts
These set the new explicit XFB members on nir_variable.
This is needed to support ARB_gl_spirv, as Vulkan doesn't support
transform feedback.
---
src/compiler/spirv/vtn_variables.c | 12
Hi Ian,
Are you able to take a look at this one? Looks like you have been
involved in this area in the past.
Thanks,
Tim
On 15/07/18 18:04, Timothy Arceri wrote:
Fixes new piglit test:
tests/spec/glsl-1.20/execution/qualifiers/vs-out-conversion-int-to-float-vec4-index.shader_test
On 20/07/18 14:53, Dave Airlie wrote> On 20 July 2018 at 13:12, Marek
Olšák wrote:
From: Marek Olšák
To make dEQP-GLES31.functional.ssbo.layout.random.all_shared_buffer.23
finish sooner on the older CPUs. (otherwise it gets killed and we fail
the test)
I think this is possibly a bad idea,
On 19.07.18 11:47, Timothy Arceri wrote:
On 19/07/18 08:31, Eric Anholt wrote:
Danylo Piliaiev writes:
After optimization passes and many trasfromations most of memory
"transformations"
NIR holds is a garbage which was being freed only after shader
deletion.
"is gar
On 19/07/18 08:31, Eric Anholt wrote:
Danylo Piliaiev writes:
After optimization passes and many trasfromations most of memory
"transformations"
NIR holds is a garbage which was being freed only after shader deletion.
"is garbage"
Freeing it at the end of linking will save memory
For simplicity we only allow this to work for scalar types.
---
src/compiler/nir/nir_search.c | 29 ++---
1 file changed, 22 insertions(+), 7 deletions(-)
diff --git a/src/compiler/nir/nir_search.c b/src/compiler/nir/nir_search.c
index 743ffdf232c..c0224ca4360 100644
---
These appear in a single Dirt Showdown compute shader. With this
VGPR spilling in the radeonsi NIR backend is reduced.
---
src/compiler/nir/nir_opt_algebraic.py | 18 ++
1 file changed, 18 insertions(+)
diff --git a/src/compiler/nir/nir_opt_algebraic.py
---
src/compiler/nir/nir_search.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/compiler/nir/nir_search.c b/src/compiler/nir/nir_search.c
index 28b36b2b863..743ffdf232c 100644
--- a/src/compiler/nir/nir_search.c
+++ b/src/compiler/nir/nir_search.c
@@ -73,6 +73,9 @@
the code required?
On Thu, Jul 19, 2018 at 8:06 AM, Timothy Arceri wrote:
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All helped shaders are from Unreal Engine 4 besides one shader from
Dirt Showdown.
V2: make sure we do evaluation when
Since we know what side of the branch we ended up on we can just
replace the use with a constant.
All helped shaders are from Unreal Engine 4 besides one shader from
Dirt Showdown.
V2: make sure we do evaluation when condition is used in else with
a single block (we were checking for blocks
On 19/07/18 03:07, Brian Paul wrote:
On 07/17/2018 06:47 PM, Stefan Dirsch wrote:
On Tue, Jul 17, 2018 at 04:57:26PM -0600, Brian Paul wrote:
Reviewed-by: Brian Paul
Do you need me to push this for you?
I'm afraid the answer is yes. Tried it but push hangs forever after this
# git push
This does two things:
1. Allows cross shader optimisations to be preformed on SSO
programs that contain more than one stage (seems like I
don't have any of these in my shader-db collection).
2. Allows us to write compilation errors to stderr. With this
change I discovered a Deus Ex: MD
Series:
Reviewed-by: Timothy Arceri
On 17/07/18 15:58, Marek Olšák wrote:
From: Marek Olšák
This should only contain functions not set in vtxfmt.c.
---
src/mesa/main/api_loopback.c | 17 -
1 file changed, 17 deletions(-)
diff --git a/src/mesa/main/api_loopback.c b/src
On 17/07/18 03:41, Eric Anholt wrote:
Timothy Arceri writes:
This makes this opt behave more like the GLSL IR opt
lower_if_to_cond_assign(). With this we can disable that GLSL IR
opt on drivers with a NIR backend without causing spill
regressions.
shader-db results for radeonsi (RX580
One of these was seen in a Deus Ex shader.
V2: make opts more generic.
Cc: Jason Ekstrand
---
src/compiler/nir/nir_opt_algebraic.py | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/compiler/nir/nir_opt_algebraic.py
b/src/compiler/nir/nir_opt_algebraic.py
index
On 16/07/18 14:25, Jason Ekstrand wrote:
How about just adding
(('ior', 'a@bool', ('ieq', a, 0)), True)
and maybe also
(('ior', 'a@bool', ('inot', a)), True)
while you're at it.
Ah yes that's much better. Will do.
--Jason
On Sun, Jul 15, 2018 at 4:27 PM Timothy Arceri <mailto:t
This allows NIR to CSE more operations. LLVM does this also so the
impact is limited, however doing this in NIR allows other opts to
make progress.
I didn't have a large vkpipeline-db to run this against but for
radeonsi enabling this allowed some loops in Civilization Beyond
Earth shaders to
This allows NIR to CSE more operations. LLVM does this also so the
impact is limited, however doing this in NIR allows other opts to
make progress. For example some loops in Civilization Beyond Earth
shaders are unrolled.
---
src/gallium/drivers/radeonsi/si_shader_nir.c | 2 ++
1 file changed, 2
This makes this opt behave more like the GLSL IR opt
lower_if_to_cond_assign(). With this we can disable that GLSL IR
opt on drivers with a NIR backend without causing spill
regressions.
shader-db results for radeonsi (RX580):
Totals from affected shaders:
SGPRS: 12200 -> 13072 (7.15 %)
VGPRS:
One of these was seen in a Deus Ex shader.
---
src/compiler/nir/nir_opt_algebraic.py | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/compiler/nir/nir_opt_algebraic.py
b/src/compiler/nir/nir_opt_algebraic.py
index 0986dde6851..682d498b79b 100644
---
Cc: Jason Ekstrand
---
src/compiler/nir/nir_opt_peephole_select.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/compiler/nir/nir_opt_peephole_select.c
b/src/compiler/nir/nir_opt_peephole_select.c
index 8fa99117f4c..ad9d0abec03 100644
--- a/src/compiler/nir/nir_opt_peephole_select.c
Noticed in a unigine shader.
---
src/compiler/nir/nir_opt_algebraic.py | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/compiler/nir/nir_opt_algebraic.py
b/src/compiler/nir/nir_opt_algebraic.py
index 5e07d662b07..0986dde6851 100644
--- a/src/compiler/nir/nir_opt_algebraic.py
+++
601 - 700 of 6672 matches
Mail list logo
