01.01.2017 06:41, Kenneth Graunke пишет:
On Sunday, January 1, 2017 1:34:27 AM PST Marek Olšák wrote:
From: Marek Olšák
This reduces compile times by 4.5% with the Gallium noop driver and
gl_constants::GLSLOptimizeConservatively == true.
Compile times of...what exactly? Do you have any sta
On Jan 1, 2017 4:41 AM, "Kenneth Graunke" wrote:
On Sunday, January 1, 2017 1:34:27 AM PST Marek Olšák wrote:
> From: Marek Olšák
>
> This reduces compile times by 4.5% with the Gallium noop driver and
> gl_constants::GLSLOptimizeConservatively == true.
Compile times of...what exactly? Do you
On Jan 1, 2017 11:55 AM, "Vladislav Egorov" wrote:
01.01.2017 06:41, Kenneth Graunke пишет:
On Sunday, January 1, 2017 1:34:27 AM PST Marek Olšák wrote:
>
>> From: Marek Olšák
>>
>> This reduces compile times by 4.5% with the Gallium noop driver and
>> gl_constants::GLSLOptimizeConservatively
https://bugs.freedesktop.org/show_bug.cgi?id=99240
Bug ID: 99240
Summary: glxinfo reports different extensions for wayland vs
xorg.
Product: Mesa
Version: 13.0
Hardware: Other
OS: All
Status: NEW
https://bugs.freedesktop.org/show_bug.cgi?id=99240
--- Comment #1 from pandiculationfi...@gmail.com ---
Created attachment 128705
--> https://bugs.freedesktop.org/attachment.cgi?id=128705&action=edit
xorg glxinfo output
--
You are receiving this mail because:
You are the QA Contact for the bug
https://bugs.freedesktop.org/show_bug.cgi?id=99240
--- Comment #2 from pandiculationfi...@gmail.com ---
diff of the first 50 or so lines of glxinfo.
diff wayland-glxinfo.txt xorg-glxinfo.txt | head -50
7,8c7,9
< GLX_ARB_create_context, GLX_ARB_create_context_profile,
< GLX_ARB_fbconfig_flo
Really, we should have some kind of function for copying the whole
table, but this will work for now.
---
src/compiler/glsl/opt_copy_propagation.cpp | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/src/compiler/glsl/opt_copy_propagation.cpp
b/src/compiler/glsl/opt_copy_pr
Hi all.
This is a series I've put together to lower the overhead of
the copy propagation pass in glsl. The series is not refined,
so please treat it as a proof of concept / RFC.
The first five patches might be merge-ready, but their benefit is
quite low compared to the actual major change here.
---
src/compiler/nir/nir_lower_locals_to_regs.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/src/compiler/nir/nir_lower_locals_to_regs.c
b/src/compiler/nir/nir_lower_locals_to_regs.c
index f1af237b5e..d0667bc504 100644
--- a/src/compiler/nir/nir_lower_locals_to_regs.c
+++ b/src/compiler/nir
We should really have a function to copy the table contents,
but this will at least get us somewhere in the meantime.
---
src/compiler/glsl/opt_copy_propagation_elements.cpp | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/src/compiler/glsl/opt_copy_propagation_elements.cp
---
src/util/non_replacing_hash_table.h | 5 +
1 file changed, 5 insertions(+)
diff --git a/src/util/non_replacing_hash_table.h
b/src/util/non_replacing_hash_table.h
index b35ee871bb..7550400539 100644
--- a/src/util/non_replacing_hash_table.h
+++ b/src/util/non_replacing_hash_table.h
@@ -42
---
src/util/Makefile.sources | 2 +
src/util/non_replacing_hash_table.c | 504
src/util/non_replacing_hash_table.h | 159
3 files changed, 665 insertions(+)
create mode 100644 src/util/non_replacing_hash_table.c
create mode 100644 sr
This should give us lower memory consumption, allocation overhead,
better cache locality, and all this nice stuff.
---
src/compiler/glsl/opt_copy_propagation.cpp | 57 +-
1 file changed, 25 insertions(+), 32 deletions(-)
diff --git a/src/compiler/glsl/opt_copy_propagat
--
We might want to merge u_vector and this array, or just replace this
implementation completely. Having two separate implementations, where
one of them has one sole user seems a bit wierd.
---
src/compiler/Makefile.sources | 1 -
src/compiler/spirv/vtn_cfg.c
---
src/util/non_replacing_hash_table.c | 9 +
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/src/util/non_replacing_hash_table.c
b/src/util/non_replacing_hash_table.c
index 33c93452c4..1128388ab8 100644
--- a/src/util/non_replacing_hash_table.c
+++ b/src/util/non_replacing
---
src/util/non_replacing_hash_table.c | 18 +-
1 file changed, 17 insertions(+), 1 deletion(-)
diff --git a/src/util/non_replacing_hash_table.c
b/src/util/non_replacing_hash_table.c
index 1128388ab8..48f42aa7c9 100644
--- a/src/util/non_replacing_hash_table.c
+++ b/src/util/non
---
src/util/non_replacing_hash_table.c | 47 -
src/util/non_replacing_hash_table.h | 35 +--
2 files changed, 1 insertion(+), 81 deletions(-)
diff --git a/src/util/non_replacing_hash_table.c
b/src/util/non_replacing_hash_table.c
index
---
src/util/non_replacing_hash_table.c | 41 +++--
1 file changed, 35 insertions(+), 6 deletions(-)
diff --git a/src/util/non_replacing_hash_table.c
b/src/util/non_replacing_hash_table.c
index 1730b1ee0d..a4c57e9349 100644
--- a/src/util/non_replacing_hash_table.
This way we can build one hash table in each direction.
This allows us to move the algorithm from O(n^2) to O(2*n).
The downside is an effective doubling of memory consumption
---
src/compiler/glsl/opt_copy_propagation.cpp | 49 +++---
1 file changed, 45 insertions(+), 4 de
---
src/util/non_replacing_hash_table.c | 6 ++
1 file changed, 6 insertions(+)
diff --git a/src/util/non_replacing_hash_table.c
b/src/util/non_replacing_hash_table.c
index a4c57e9349..33c93452c4 100644
--- a/src/util/non_replacing_hash_table.c
+++ b/src/util/non_replacing_hash_table.c
@@ -1
---
src/util/non_replacing_hash_table.c | 112 +++-
src/util/non_replacing_hash_table.h | 102
2 files changed, 111 insertions(+), 103 deletions(-)
diff --git a/src/util/non_replacing_hash_table.c
b/src/util/non_replacing_hash_tabl
Oh, and an important disclaimer.
I have not ran this through picking, nor have I done a complete
run of shader-db for performance comparisons on more realistic workloads.
I'll see what time I can find this evening, or some time tomorrow.
On Jan 1, 2017 19:38, "Thomas Helland" wrote:
Hi all.
Thi
On Sat, Dec 31, 2016 at 1:31 PM, Romain Failliot
wrote:
> I tried to use git send-email but it doesn't seem to work (although
> the output says otherwise).
>
> So eventually it's simpler to just copy/paste the patch generated by
> git format-patch:
well, I was trying to not loose authorship/etc f
It's more user friendly and it avoids to write files in unexpected
places.
---
src/gallium/auxiliary/hud/hud_context.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/gallium/auxiliary/hud/hud_context.c
b/src/gallium/auxiliary/hud/hud_context.c
index 779c116..cccdb26 100
Darn it. There appears to be a functional change with patch 14.
My changes where meant to make no functional change,
so something is fishy. An example from the public shader-db:
cycles HURT: ./shaders/tesseract/205.shader_test VS vec4: 54 ->
100 (85.19%)
I'll see if I can find some time to
On 16-12-10 16:05:12, Pohjolainen, Topi wrote:
On Thu, Dec 01, 2016 at 02:10:01PM -0800, Ben Widawsky wrote:
From: Ben Widawsky
Signed-off-by: Ben Widawsky
---
src/mesa/drivers/dri/i965/intel_screen.c | 20 +++-
1 file changed, 15 insertions(+), 5 deletions(-)
diff --git a/s
So I've discovered the issue. I wasn't inserting into both tables.
So we never did any copy propagation. So that was stupid.
Now it assertion fails in _insert_pre_hashed as hashing the
key does not return the same result as the supplied hash.
I'll look more into this some time during the week.
201
On Sun, 2017-01-01 at 16:15 -0500, Rob Clark wrote:
> On Sat, Dec 31, 2016 at 1:31 PM, Romain Failliot
> wrote:
> > I tried to use git send-email but it doesn't seem to work (although
> > the output says otherwise).
> >
> > So eventually it's simpler to just copy/paste the patch generated by
> >
2017-01-01 23:20 GMT+01:00 Thomas Helland :
> So I've discovered the issue. I wasn't inserting into both tables.
> So we never did any copy propagation. So that was stupid.
> Now it assertion fails in _insert_pre_hashed as hashing the
> key does not return the same result as the supplied hash.
> I'
On 16-12-10 16:24:46, Pohjolainen, Topi wrote:
On Thu, Dec 01, 2016 at 02:10:06PM -0800, Ben Widawsky wrote:
From: Ben Widawsky
On Gen9 hardware, the display engine is able to scanout a compressed
framebuffer by providing an offset to auxiliary compression information.
Unfortunately, the hardw
On 16-12-11 10:05:25, Pohjolainen, Topi wrote:
On Thu, Dec 01, 2016 at 02:10:07PM -0800, Ben Widawsky wrote:
From: Ben Widawsky
Cc: Topi Pohjolainen
Signed-off-by: Ben Widawsky
---
src/mesa/drivers/dri/i965/intel_mipmap_tree.c | 12 +++-
1 file changed, 3 insertions(+), 9 deletions(
As CCS buffers are passed to KMS, it becomes useful to be able to
determine exactly what type of aux buffers are disabled. This was
previously not entirely needed (though the code was a little more
confusing), however it becomes very desirable after a recent patch from
Chad:
commit 1c8be049bea786c
Cc: Topi Pohjolainen
Cc: Chad Versace
Signed-off-by: Ben Widawsky
---
src/mesa/drivers/dri/i965/brw_blorp.c | 2 +-
src/mesa/drivers/dri/i965/brw_draw.c | 3 ++-
src/mesa/drivers/dri/i965/intel_mipmap_tree.c | 21 ++---
src/mesa/drivers/dri/i965/intel_mipmap_t
https://bugs.freedesktop.org/show_bug.cgi?id=99244
Bug ID: 99244
Summary: hud_context.c:874:45: error: ‘W_OK’ undeclared
Product: Mesa
Version: git
Hardware: x86-64 (AMD64)
OS: Linux (All)
Status: NEW
Keyw
Signed-off-by: Ilia Mirkin
---
src/compiler/glsl/lower_output_reads.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/compiler/glsl/lower_output_reads.cpp
b/src/compiler/glsl/lower_output_reads.cpp
index 851078b..bd3accb 100644
--- a/src/compiler/glsl/lower_output_reads
This implements support for emitting FBFETCH ops, using the existing
lowering pass for advanced blend logic, and disabling hw blend when
advanced blending is enabled.
Signed-off-by: Ilia Mirkin
---
src/mesa/state_tracker/st_atom_blend.c| 2 +-
src/mesa/state_tracker/st_cb_texturebarrier
This is so that we can differentiate between flushing any framebuffer
reading caches from regular sampler caches.
Signed-off-by: Ilia Mirkin
---
This felt too simple and silly to create an extra callback for, especially
since the implementations that rely on texture sampling to retrieve fb
conte
The extension spec is not currently published, so it's a bit premature
to require it for BlendBarrier usage.
Signed-off-by: Ilia Mirkin
---
src/mesa/main/barrier.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/mesa/main/barrier.c b/src/mesa/main/barrier.c
index 2f5b45
Signed-off-by: Ilia Mirkin
---
src/gallium/docs/source/screen.rst | 2 ++
src/gallium/drivers/freedreno/freedreno_screen.c | 1 +
src/gallium/drivers/i915/i915_screen.c | 1 +
src/gallium/drivers/ilo/ilo_screen.c | 1 +
src/gallium/drivers/llvmpipe/lp_screen.c
Signed-off-by: Ilia Mirkin
---
src/gallium/auxiliary/tgsi/tgsi_info.c | 2 +-
src/gallium/docs/source/tgsi.rst | 11 +++
src/gallium/include/pipe/p_shader_tokens.h | 2 +-
3 files changed, 13 insertions(+), 2 deletions(-)
diff --git a/src/gallium/auxiliary/tgsi/tgsi_info.
We don't need to support all the color buffers for advanced blend, just
cb0. For Fermi, we use the special binding slots so that we don't
overlap with user textures, while Kepler+ gets a dedicated position for
the fb handle in the driver constbuf.
This logic is only triggered when a FBREAD is actu
On Sat, Dec 31, 2016 at 02:40:42PM -0800, Ben Widawsky wrote:
> On 16-12-10 15:39:12, Pohjolainen, Topi wrote:
> > On Thu, Dec 01, 2016 at 02:09:59PM -0800, Ben Widawsky wrote:
> > > From: Ben Widawsky
> > >
> > > There is nothing particularly useful to do currently if the update
> > > fails, but
42 matches
Mail list logo