On 22/08/17 10:48 AM, Timothy Arceri wrote:
> This reverts commit ca73c3358c91434e68ab31c23d13986d87c661cd.
>
> This commit broke 'make check' revert it for now.
Acked-by: Michel Dänzer
--
Earthling Michel Dänzer | http://www.amd.com
Libre
Reviewed-by: Samuel Iglesias Gonsálvez
On Mon, 2017-08-21 at 22:11 -0700, Jason Ekstrand wrote:
> I have no idea how this got missed but it's been missing since
> forever.
>
> Cc: mesa-sta...@lists.freedesktop.org
> ---
> src/compiler/spirv/vtn_variables.c | 5 -
> 1
https://bugs.freedesktop.org/show_bug.cgi?id=102052
Vinson Lee changed:
What|Removed |Added
Resolution|--- |FIXED
I have no idea how this got missed but it's been missing since forever.
Cc: mesa-sta...@lists.freedesktop.org
---
src/compiler/spirv/vtn_variables.c | 5 -
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/src/compiler/spirv/vtn_variables.c
b/src/compiler/spirv/vtn_variables.c
On Saturday, August 19, 2017 8:56:06 PM PDT Jason Ekstrand wrote:
> In f9fd976e8adba733b08d we changed the clear value to be stored as an
> isl_color_value. This had the side-effect same clear value check is now
> happening directly between the f32[0] field of the isl_color_value and
>
> -Original Message-
> From: mesa-dev [mailto:mesa-dev-boun...@lists.freedesktop.org] On Behalf
> Of Emil Velikov
> Sent: Monday, August 21, 2017 11:24 PM
> To: Marathe, Yogesh
> Cc: Gao, Shuo ; Liu, Zhiquan ;
>
On Mon, Aug 21, 2017 at 9:57 PM, Ian Romanick wrote:
> From: Ian Romanick
>
> copy_constant_to_storage, set_uniform_initializer,
> populate_consumer_input_sets, and get_matching_input are all used by
> tests in src/compiler/glsl/tests:
Reviewed-by: Timothy Arceri
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
From: Ian Romanick
copy_constant_to_storage, set_uniform_initializer,
populate_consumer_input_sets, and get_matching_input are all used by
tests in src/compiler/glsl/tests:
glsl/tests/varyings_test.o: In function
`link_varyings_single_simple_input_Test::TestBody()':
This reverts commit ca73c3358c91434e68ab31c23d13986d87c661cd.
This commit broke 'make check' revert it for now.
---
src/compiler/glsl/ast_function.cpp | 10 +-
src/compiler/glsl/ast_to_hir.cpp| 6 +++---
src/compiler/glsl/glsl_parser_extras.cpp| 2
On 08/18/2017 04:02 AM, Nicolai Hähnle wrote:
On 15.08.2017 01:26, Timothy Arceri wrote:
This will be used to store more information about the cache item
in it's header. This information is intended for 3rd party and
cache analysis use but can also be used for detecting the unlikely
scenario
Hi Jason,
-Ursprüngliche Nachricht-
Von: mesa-dev [mailto:mesa-dev-boun...@lists.freedesktop.org] Im Auftrag von
Jason Ekstrand
Gesendet: Donnerstag, 17. August 2017 23:57
An: mesa-dev@lists.freedesktop.org
Cc: Jason Ekstrand
Betreff: [Mesa-dev] [PATCH 2/2]
Hi Julien,
thanks for providing some background on the issue. Now it makes a lot
more sense.
Am Mittwoch, den 16.08.2017, 10:40 +0100 schrieb Julien Isorce:
> Hi,
>
> This issue is tracked here https://bugs.freedesktop.org/show_bug.cgi?
> id=98285
> This is due to a limitation in libva API
On Mon, Aug 21, 2017 at 2:44 PM, Matt Turner wrote:
> On Thu, Aug 10, 2017 at 7:02 PM, Jordan Justen
> wrote:
> > On 2017-08-10 15:02:33, Matt Turner wrote:
> >> Quiets a number of uninitialized variable warnings in clang.
> >> ---
> >>
https://bugs.freedesktop.org/show_bug.cgi?id=102038
--- Comment #11 from Bruce Cherniak ---
Created attachment 133661
--> https://bugs.freedesktop.org/attachment.cgi?id=133661=edit
simple-msaa.c
Hi Brian,
This simple msaa sample fails in the same way when run with
https://bugs.freedesktop.org/show_bug.cgi?id=102038
--- Comment #10 from Bruce Cherniak ---
For llvmpipe, this fails in the same way as before. Unless
llvmpipe_is_format_supported is changed in the same way that you modified
svga_is_format_supported:
- if
This one is not needed, I was testing with missing pieces in
external/llvm on my side
Sorry for the noise
Mauro
2017-08-21 23:49 GMT+02:00 Mauro Rossi :
> Fixes the following building errors:
>
> In file included from external/mesa/src/amd/common/ac_shader_info.c:25:
> In
2017-08-21 23:57 GMT+02:00 Rob Herring :
> On Mon, Aug 21, 2017 at 4:44 PM, Mauro Rossi wrote:
>> 2017-08-18 21:46 GMT+02:00 Rob Herring :
>>> Both statically linking libLLVMCore and dynamically linking libLLVM causes
>>> duplicated symbols
Am 21.08.2017 um 22:57 schrieb Marek Olšák:
> On Sun, Aug 20, 2017 at 4:29 AM, Roland Scheidegger
> wrote:
>> For 1-3/8, I don't think anything ever used these, and 5/8 (DPH) we
>> don't need neither, so
>> Reviewed-by: Roland Scheidegger
>>
>> 4,6,7,
On Mon, Aug 21, 2017 at 4:44 PM, Mauro Rossi wrote:
> 2017-08-18 21:46 GMT+02:00 Rob Herring :
>> Both statically linking libLLVMCore and dynamically linking libLLVM causes
>> duplicated symbols in gallium_dri.so and it fails to dlopen. We don't
>> really
On Mon, Aug 21, 2017 at 5:54 PM, Andy Furniss wrote:
> Ilia Mirkin wrote:
>
>> As I don't have the hardware myself, I was hoping someone could
>> confirm that this is a nouveau issue and not a more general one.
>> Ideally this would be done by testing VDPAU on both radeonsi
From: Marek Olšák
+ 4 piglit regressions, but it's correct accorcing to the GL spec and
performance is more important than piglit.
---
src/gallium/drivers/radeonsi/si_shader_tgsi_alu.c | 29 +++
1 file changed, 8 insertions(+), 21 deletions(-)
diff
From: Marek Olšák
---
src/gallium/drivers/radeonsi/si_blit.c | 27 +++
1 file changed, 15 insertions(+), 12 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_blit.c
b/src/gallium/drivers/radeonsi/si_blit.c
index 378441a..2544388 100644
---
From: Marek Olšák
---
src/gallium/drivers/radeonsi/si_pipe.c| 2 +-
src/gallium/drivers/radeonsi/si_shader_tgsi_alu.c | 14 ++
2 files changed, 15 insertions(+), 1 deletion(-)
diff --git a/src/gallium/drivers/radeonsi/si_pipe.c
From: Marek Olšák
---
src/gallium/drivers/r600/evergreen_state.c| 3 +--
src/gallium/drivers/r600/r600_blit.c | 3 +--
src/gallium/drivers/r600/r600_state.c | 3 +--
src/gallium/drivers/radeon/r600_pipe_common.h | 13 +
From: Marek Olšák
clip_regs aren't marked dirty when writes_viewport_index is changed.
Cc: 17.2
---
src/gallium/drivers/radeonsi/si_state.c | 6 --
src/gallium/drivers/radeonsi/si_state_shaders.c | 11 +--
2 files
From: Marek Olšák
---
src/gallium/drivers/radeonsi/si_blit.c | 12 +---
src/gallium/drivers/radeonsi/si_pipe.h | 17 ++---
src/gallium/drivers/radeonsi/si_state.c | 8 +++-
3 files changed, 26 insertions(+), 11 deletions(-)
diff --git
From: Marek Olšák
---
src/gallium/drivers/radeonsi/si_blit.c | 8 +---
src/gallium/drivers/radeonsi/si_pipe.h | 23 +++
src/gallium/drivers/radeonsi/si_state.c | 19 +++
src/gallium/drivers/radeonsi/si_state_draw.c |
From: Marek Olšák
Now we can finally stop flushing L2 data.
---
src/gallium/drivers/radeonsi/si_state_draw.c | 28 +---
1 file changed, 21 insertions(+), 7 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_state_draw.c
From: Marek Olšák
Vulkan doesn't do it anymore.
---
src/gallium/drivers/radeonsi/si_state_shaders.c | 10 +++---
1 file changed, 3 insertions(+), 7 deletions(-)
diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c
From: Marek Olšák
0 is PIPE_QUERY_OCCLUSION_COUNTER, which is not what we want.
---
src/gallium/drivers/radeon/r600_pipe_common.h | 2 ++
src/gallium/drivers/radeonsi/si_perfcounter.c | 2 +-
src/gallium/drivers/radeonsi/si_state_draw.c | 4 ++--
3 files changed, 5
From: Marek Olšák
---
src/amd/common/r600d_common.h | 10 --
src/gallium/drivers/radeon/r600_query.c | 9 ++---
src/gallium/drivers/radeonsi/si_perfcounter.c | 3 ++-
src/gallium/drivers/radeonsi/si_state_draw.c | 6 --
4 files
From: Marek Olšák
---
src/gallium/drivers/radeonsi/si_blit.c | 29 -
src/gallium/drivers/radeonsi/si_pipe.h | 23 +++
src/gallium/drivers/radeonsi/si_state.c | 29 -
3 files changed, 47
From: Marek Olšák
This fixes garbage there if we don't flush TC L2 after rendering.
---
src/amd/common/ac_surface.c | 19 ++-
1 file changed, 18 insertions(+), 1 deletion(-)
diff --git a/src/amd/common/ac_surface.c b/src/amd/common/ac_surface.c
index
From: Marek Olšák
- don't precompile LS and ES (they don't exist on GFX9), compile as VS instead
- don't precompile HS and GS (we don't have LS and ES parts)
---
src/gallium/drivers/radeonsi/si_state_shaders.c | 12 +++-
1 file changed, 11 insertions(+), 1
From: Marek Olšák
Cc: 17.2
---
src/amd/common/ac_surface.c | 5 +
src/gallium/drivers/radeon/r600_texture.c | 9 +++--
2 files changed, 8 insertions(+), 6 deletions(-)
diff --git a/src/amd/common/ac_surface.c
Ilia Mirkin wrote:
As I don't have the hardware myself, I was hoping someone could
confirm that this is a nouveau issue and not a more general one.
Ideally this would be done by testing VDPAU on both radeonsi and r600,
as well as both the DRI3 and the DRI2 paths. Please use mplayer for
this,
From: Marek Olšák
An island of beauty in the middle of chaos.
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 92 ++
1 file changed, 30 insertions(+), 62 deletions(-)
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
For patches 1-3, 6, 7:
Reviewed-by: Marek Olšák
Marek
On Thu, Aug 17, 2017 at 1:03 PM, Timothy Arceri wrote:
> Will be used to add LOAD support to UBOs.
> ---
> src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 46
> ++
> 1
This will use vector loads (64 loads per 64 threads) instead of scalar
loads (1 load per 64 threads). Vector loads are much slower.
allow_smem should be set to allow scalar loads for constbufs.
Also, can_speculate should be true for constbuf loads.
I think you really need to check shader-db or
Fixes the following building errors:
In file included from external/mesa/src/amd/common/ac_shader_info.c:25:
In file included from external/mesa/src/amd/common/ac_nir_to_llvm.h:29:
In file included from external/llvm/include/llvm-c/TargetMachine.h:23:
external/llvm/include/llvm-c/Target.h:23:10:
On Thu, Aug 10, 2017 at 7:02 PM, Jordan Justen
wrote:
> On 2017-08-10 15:02:33, Matt Turner wrote:
>> Quiets a number of uninitialized variable warnings in clang.
>> ---
>> src/compiler/spirv/spirv_to_nir.c | 24
>>
2017-08-18 21:46 GMT+02:00 Rob Herring :
> Both statically linking libLLVMCore and dynamically linking libLLVM causes
> duplicated symbols in gallium_dri.so and it fails to dlopen. We don't
> really need to link libLLVMCore, but just need generated headers to be
> built first.
I'd like the new "file" type to be called CONSTBUF instead of UBO. We
don't generally use GL nomenclature in Gallium.
Marek
On Thu, Aug 17, 2017 at 1:03 PM, Timothy Arceri wrote:
> This will be use to distinguish between load types when using
> the TGSI_OPCODE_LOAD
Reviewed-by: Marek Olšák
Marek
On Mon, Aug 21, 2017 at 2:42 AM, Timothy Arceri wrote:
> This code was separated from the validation code so it could
> use used with KHR_no_error paths. The return values were inverted
> to reflect the name of the
For the series:
Reviewed-by: Marek Olšák
Marek
On Mon, Aug 21, 2017 at 8:56 PM, Brian Paul wrote:
> After the context is initialized, the API and context flags won't
> change. So, we can compute whether vertex attribute 0 aliases
> vertex position just
Reviewed-by: Charmaine Lee
From: Brian Paul
Sent: Monday, August 21, 2017 11:55 AM
To: mesa-dev@lists.freedesktop.org
Cc: Charmaine Lee; Neha Bhende
Subject: [PATCH] svga: replace gotos with conditionals in
For the series:
Reviewed-by: Marek Olšák
Marek
On Mon, Aug 21, 2017 at 10:22 PM, Samuel Pitoiset
wrote:
> They should not be exposed when the extension is unsupported.
> Note that ARB_direct_state_access is always exposed and
> EXT_semaphore is
Reviewed-by: Marek Olšák
So let's commit this!
Marek
On Mon, Aug 21, 2017 at 4:50 PM, Samuel Pitoiset
wrote:
> Currently, when the array is full it is resized but it can grow
> over and over because we don't try to re-use descriptor slots.
>
>
Reviewed-by: Marek Olšák
Marek
On Mon, Aug 21, 2017 at 4:49 PM, Samuel Pitoiset
wrote:
> Will be used for allocating bindless descriptor slots for
> RadeonSI.
>
> v3; - use a faster loop in util_idalloc_alloc() (Marek)
> v2: - change the
From: Dave Airlie
---
include/vulkan/vulkan.h| 19 +
src/vulkan/registry/vk.xml | 53 +++---
2 files changed, 60 insertions(+), 12 deletions(-)
diff --git a/include/vulkan/vulkan.h b/include/vulkan/vulkan.h
index
On Sun, Aug 20, 2017 at 4:29 AM, Roland Scheidegger wrote:
> For 1-3/8, I don't think anything ever used these, and 5/8 (DPH) we
> don't need neither, so
> Reviewed-by: Roland Scheidegger
>
> 4,6,7, well we can live fine without them, just translating them
On Mon, Aug 21, 2017 at 8:49 PM, Samuel Pitoiset
wrote:
> dirty_mask can be removed entirely because it only makes sense with CE.
Thanks. Removed locally.
Marek
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
From the EXT_external_objects_fd spec:
"If the GL_EXT_memory_object_fd string is reported, the following
commands are added:
void ImportMemoryFdEXT(uint memory,
uint64 size,
enum handleType,
int fd);"
They should not be exposed when the extension is unsupported.
Note that ARB_direct_state_access is always exposed and
EXT_semaphore is not supported at all.
Signed-off-by: Samuel Pitoiset
---
src/mesa/main/bufferobj.c | 21 +++--
https://bugs.freedesktop.org/show_bug.cgi?id=102038
--- Comment #9 from Brian Paul ---
Created attachment 133654
--> https://bugs.freedesktop.org/attachment.cgi?id=133654=edit
proposed patch
Bruce, can you try the attached patch?
--
You are receiving this mail because:
On 21 August 2017 at 23:21, Marek Olšák wrote:
> I'll push your patch when I push my series. I guess that's all, right?
If anything else falls out of r600 I'll take a look, but piglit seemed fine.
(as fine as piglit ever is on evergreen).
Nice to see some radv optimisation
On 21 August 2017 at 17:44, Bas Nieuwenhuizen wrote:
> The old code would just do a * blocksize / blocksize , which is the
> identity (as long as we don't overflow). Did this cause any issues?
Nope, just seemed like unnecessary cpu usage.
Dave.
On 08/18/2017 10:25 AM, Marek Olšák wrote:
On Fri, Aug 18, 2017 at 5:56 PM, Brian Paul wrote:
On 08/17/2017 12:31 PM, Marek Olšák wrote:
Hi,
Major u_blitter changes:
- All draw calls except cubemap blits will now use the draw_rectangle
callback. (cubemap blits are
On 21 August 2017 at 23:44, Emil Velikov wrote:
> On 21 August 2017 at 14:19, Marek Olšák wrote:
>> On Mon, Aug 21, 2017 at 12:47 PM, Emil Velikov
>> wrote:
>>> Hi Marek,
>>>
>>> On 20 August 2017 at 01:49, Marek Olšák
https://bugs.freedesktop.org/show_bug.cgi?id=102038
--- Comment #8 from Bruce Cherniak ---
Roland, you are correct, VTK is trying to use msaa surfaces with just one
sample.
Here's what's happening:
The rendertarget color buffer is being created by:
st_api_make_current
Pass the dri.sym version script to the linker. This ensures only
explicitly exported symbols are exported and shrinks the library by up
to 60KB.
We need to pass "--undefined-version" because the Android build system
sets --no-undefined-version by default and we get an error on
On 08/21/2017 01:09 PM, Samuel Pitoiset wrote:
On 08/21/2017 08:56 PM, Brian Paul wrote:
After the context is initialized, the API and context flags won't
change. So, we can compute whether vertex attribute 0 aliases
vertex position just once.
This should make the glVertexAttrib*()
On 08/21/2017 08:56 PM, Brian Paul wrote:
After the context is initialized, the API and context flags won't
change. So, we can compute whether vertex attribute 0 aliases
vertex position just once.
This should make the glVertexAttrib*() functions a little quicker.
No Piglit regressions with
After the context is initialized, the API and context flags won't
change. So, we can compute whether vertex attribute 0 aliases
vertex position just once.
This should make the glVertexAttrib*() functions a little quicker.
No Piglit regressions with llvmpipe.
---
src/mesa/main/context.c | 17
---
src/mesa/vbo/vbo.h| 2 --
src/mesa/vbo/vbo_exec_array.c | 2 +-
2 files changed, 1 insertion(+), 3 deletions(-)
diff --git a/src/mesa/vbo/vbo.h b/src/mesa/vbo/vbo.h
index c8e87d3..ffbb201 100644
--- a/src/mesa/vbo/vbo.h
+++ b/src/mesa/vbo/vbo.h
@@ -196,8 +196,6 @@ void
Makes the code a bit more understandable.
---
src/mesa/vbo/vbo_attrib_tmp.h | 82 ---
1 file changed, 46 insertions(+), 36 deletions(-)
diff --git a/src/mesa/vbo/vbo_attrib_tmp.h b/src/mesa/vbo/vbo_attrib_tmp.h
index 8328445..5718ac5 100644
---
No Piglit regressions.
---
src/gallium/drivers/svga/svga_draw_arrays.c | 36 +--
src/gallium/drivers/svga/svga_draw_elements.c | 28 ++---
2 files changed, 32 insertions(+), 32 deletions(-)
diff --git a/src/gallium/drivers/svga/svga_draw_arrays.c
dirty_mask can be removed entirely because it only makes sense with CE.
On 08/19/2017 07:03 PM, Marek Olšák wrote:
From: Marek Olšák
We have come to the conclusion that it doesn't improve performance.
---
src/gallium/drivers/radeon/r600_gpu_load.c| 3 -
On Mon, Aug 21, 2017 at 12:47 PM, Emil Velikov wrote:
> On 21 August 2017 at 17:17, Rob Herring wrote:
>> On Fri, Aug 18, 2017 at 7:27 PM, Emil Velikov
>> wrote:
>>> On 18 August 2017 at 20:46, Rob Herring
Signed-off-by: Leo Liu
---
src/gallium/drivers/radeon/radeon_vcn_dec.c | 18 +++---
1 file changed, 11 insertions(+), 7 deletions(-)
diff --git a/src/gallium/drivers/radeon/radeon_vcn_dec.c
b/src/gallium/drivers/radeon/radeon_vcn_dec.c
index 51391627d5..a7a57f2969
Signed-off-by: Leo Liu
---
src/gallium/state_trackers/va/picture.c | 12
1 file changed, 12 insertions(+)
diff --git a/src/gallium/state_trackers/va/picture.c
b/src/gallium/state_trackers/va/picture.c
index e0287d30f2..b2be7af8c4 100644
---
On 21 August 2017 at 19:12, Eric Anholt wrote:
> Emil Velikov writes:
>
>> On 17 August 2017 at 17:10, Eric Anholt wrote:
>>> This is how VC4 stores 5551 textures, which we need to support for
>>> GL_OES_required_internalformat.
>>>
since the way should be as same as UVD
Signed-off-by: Leo Liu
---
src/gallium/drivers/radeon/radeon_vcn_dec.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/drivers/radeon/radeon_vcn_dec.c
b/src/gallium/drivers/radeon/radeon_vcn_dec.c
index
Emil Velikov writes:
> On 17 August 2017 at 17:10, Eric Anholt wrote:
>> This is how VC4 stores 5551 textures, which we need to support for
>> GL_OES_required_internalformat.
>>
>> v2: Extend commit message, fix svga driver build, add BE ordering from
On 21 August 2017 at 16:45, Marathe, Yogesh wrote:
> Can someone please comment on this? Are we done / still not done here?
>
I would have s/deinit/fini/ but we can squash that on merge or at a later point.
Regardless of the above:
Reviewed-by: Emil Velikov
On 21 August 2017 at 17:17, Rob Herring wrote:
> On Fri, Aug 18, 2017 at 7:27 PM, Emil Velikov
> wrote:
>> On 18 August 2017 at 20:46, Rob Herring wrote:
>>> Both statically linking libLLVMCore and dynamically linking libLLVM causes
On 21 August 2017 at 17:26, Rob Herring wrote:
> On Mon, Aug 21, 2017 at 9:08 AM, Emil Velikov
> wrote:
>> As said in the EGL_KHR_platform_android extensions
>>
>> For each EGLConfig that belongs to the Android platform, the
>>
On 21 August 2017 at 15:44, Daniel Stone wrote:
> Hi Emil,
>
> On 21 August 2017 at 15:18, Emil Velikov wrote:
>> On 11 July 2017 at 14:27, Emil Velikov wrote:
>>> According to the EGL_KHR_platform_gbm extension:
>>>
>>>
On Monday, 2017-08-21 18:09:01 +0100, Emil Velikov wrote:
> On 21 August 2017 at 14:33, Eric Engestrom wrote:
> > On Monday, 2017-08-21 11:27:20 +0100, Emil Velikov wrote:
> >> From: Emil Velikov
> >>
> >> Flesh the duplicated functionality
Looks like you no longer encode shader_info::stage (which is the first
field), is that expected?
On 08/14/2017 01:49 AM, Timothy Arceri wrote:
This is so we always create reproducible cache entries. Consistency
is required for verification of any third party distributed shaders.
---
On 21 August 2017 at 14:33, Eric Engestrom wrote:
> On Monday, 2017-08-21 11:27:20 +0100, Emil Velikov wrote:
>> From: Emil Velikov
>>
>> Flesh the duplicated functionality to a helper.
>> Use const char * + sscanf to keep the compiler
On 21 August 2017 at 14:27, Eric Engestrom wrote:
> On Monday, 2017-08-21 11:27:19 +0100, Emil Velikov wrote:
>> From: Emil Velikov
>>
>> The same declaration was being used for two distinct things - a short
>> list of EGL and the core GL
My main concern with a todo like this is that it will probably be
unmaintained at some point. :)
On 08/21/2017 04:46 AM, Timothy Arceri wrote:
Generally I'm not a huge fan of todo.txt files as they tend to not
get updated. However we have a faily active community looking for
ways to contribute
Patches 2,4-5,7 are:
Reviewed-by: Samuel Pitoiset
On 08/21/2017 04:46 AM, Timothy Arceri wrote:
Outdated, features.txt is used instead.
---
It would be great if someone who knows how could just remove the wiki
page. It doesn't really offer anything of value
On Mon, Aug 21, 2017 at 9:08 AM, Emil Velikov wrote:
> As said in the EGL_KHR_platform_android extensions
>
> For each EGLConfig that belongs to the Android platform, the
> EGL_NATIVE_VISUAL_ID attribute is an Android window format, such as
>
On Fri, Aug 18, 2017 at 7:27 PM, Emil Velikov wrote:
> On 18 August 2017 at 20:46, Rob Herring wrote:
>> Both statically linking libLLVMCore and dynamically linking libLLVM causes
>> duplicated symbols in gallium_dri.so and it fails to dlopen. We don't
On Monday, 2017-08-21 17:06:19 +0200, Samuel Pitoiset wrote:
> On 08/21/2017 04:59 PM, Eric Engestrom wrote:
> > On Monday, 2017-08-21 16:49:48 +0200, Samuel Pitoiset wrote:
> > > Will be used for allocating bindless descriptor slots for
> > > RadeonSI.
> > >
> > > v3; - use a faster loop in
Can someone please comment on this? Are we done / still not done here?
Regards,
Yogesh.
> -Original Message-
> From: Marathe, Yogesh
> Sent: Friday, August 18, 2017 3:39 PM
> To: mesa-dev@lists.freedesktop.org
> Cc: tf...@chromium.org; emil.l.veli...@gmail.com; Gao, Shuo
>
Mesa 17.1.7 is now available.
In this release we have:
The state tracker received a fix to avoid a crash accessing a null
pointer exposed using llvmpipe on Windows.
The GLSL compiler has received a fix that prevents an assertion in an
error code path with invalid types.
Intel drivers have a
Series is:
Reviewed-by: Samuel Pitoiset
On 08/16/2017 01:13 PM, Nicolai Hähnle wrote:
Hi all,
the first 3 are a bunch of CE-related changes that I still had flying
around from our experiments. Patches 1+3 are clear improvements, 2 is
a wash with how we're currently
On 08/21/2017 04:59 PM, Eric Engestrom wrote:
On Monday, 2017-08-21 16:49:48 +0200, Samuel Pitoiset wrote:
Will be used for allocating bindless descriptor slots for
RadeonSI.
v3; - use a faster loop in util_idalloc_alloc() (Marek)
v2: - change the interface (remove lock/unlock)
- make a
On Monday, 2017-08-21 16:49:48 +0200, Samuel Pitoiset wrote:
> Will be used for allocating bindless descriptor slots for
> RadeonSI.
>
> v3; - use a faster loop in util_idalloc_alloc() (Marek)
> v2: - change the interface (remove lock/unlock)
> - make a non-inline implementation
>
>
Currently, when the array is full it is resized but it can grow
over and over because we don't try to re-use descriptor slots.
v4: - rebase on top of idalloc changes
v3: - use new idalloc gallium module
Signed-off-by: Samuel Pitoiset
---
Using VRAM address as bindless handles is not a good idea because
we have to use LLVMIntToPTr and the LLVM CSE pass can't optimize
because it has no information about the pointer.
Instead, use slots indexes like the existing descriptors. Note
that we use fixed 16-dword slots for both samplers and
To share common code between rw buffers and bindless descriptors.
v3: - rename to si_emit_global_shader_pointers()
Signed-off-by: Samuel Pitoiset
Reviewed-by: Marek Olšák (v3)
---
src/gallium/drivers/radeonsi/si_descriptors.c | 57
A new pair of user SGPR is needed for loading the bindless
descriptors from shaders. Because the descriptors are global for
all stages, there is no need to add separate indices for GFX9.
v4: - do not remove those unused inputs
v3: - fix merged shaders on GFX9
v2: - fix declaring new bindless
Looks like it's useless to initialize that field when CE is
unused. This will also allow to declare more than 64 elements
for the array of bindless descriptors.
Signed-off-by: Samuel Pitoiset
Reviewed-by: Marek Olšák
---
The number of bindless descriptors is dynamic and we definitely
have to support more than 256 slots.
Signed-off-by: Samuel Pitoiset
Reviewed-by: Marek Olšák
---
src/gallium/drivers/radeonsi/si_state.h | 10 +-
1 file changed, 5
Will be used for allocating bindless descriptor slots for
RadeonSI.
v3; - use a faster loop in util_idalloc_alloc() (Marek)
v2: - change the interface (remove lock/unlock)
- make a non-inline implementation
Signed-off-by: Samuel Pitoiset
Reviewed-by: Marek Olšák
1 - 100 of 147 matches
Mail list logo