https://bugs.freedesktop.org/show_bug.cgi?id=92054
Vinson Lee changed:
What|Removed |Added
Keywords||bisected
Mesa 10.6.8 is now available.
With this release we have mostly nouveau fixes, although we have the odd
i965, llvmpipe (big endian) and gbm bugfix.
Alejandro Piñeiro (1):
i965/vec4: fill src_reg type using the constructor type parameter
Antia Puentes (1):
i965/vec4: Fix saturation
---
src/glsl/opt_constant_propagation.cpp | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/src/glsl/opt_constant_propagation.cpp
b/src/glsl/opt_constant_propagation.cpp
index b6cbf61..272a180 100644
--- a/src/glsl/opt_constant_propagation.cpp
+++
Kenneth Graunke in 4654439fdd766f79a78fe0d812fd916f5815e7e6 refactored
kill sets in opt_constant_propagation() to use hash tables. This patch set
makes the suggested changes to use hash tables for acp sets, as well as
correcting documentation within the file for recent hash table changes.
RFC as
---
src/glsl/opt_constant_propagation.cpp | 63 +--
1 file changed, 38 insertions(+), 25 deletions(-)
diff --git a/src/glsl/opt_constant_propagation.cpp
b/src/glsl/opt_constant_propagation.cpp
index 184aaa1..b6cbf61 100644
---
https://bugs.freedesktop.org/show_bug.cgi?id=92052
Bug ID: 92052
Summary: nir/nir_builder.h:79: error: expected
primary-expression before ‘.’ token
Product: Mesa
Version: git
Hardware: x86-64 (AMD64)
OS:
s/lcd2dms/ld2dms/g in various places in this patch and others.
On Fri, Sep 18, 2015 at 4:00 AM, Neil Roberts wrote:
> In order to support 16x MSAA, skl+ has a wider version of lcd2dms that
> takes two parameters for the MCS data. The MCS data in the response
> still fits
https://bugs.freedesktop.org/show_bug.cgi?id=92054
Bug ID: 92054
Summary: make check gbm-symbols-check regression
Product: Mesa
Version: git
Hardware: x86-64 (AMD64)
OS: Linux (All)
Status: NEW
Keywords:
https://bugs.freedesktop.org/show_bug.cgi?id=92052
Vinson Lee changed:
What|Removed |Added
Keywords||bisected
Current GLSL badly optimizes the code making it incompatible with the
GL_ARB_separate_shader_objects extension.
Typical example of the current behavior:
VS:
out SHADER
{
vec4 var_zero; // location will always be 0
vec2 var_one; // location will always be 1
} VSout;
FS:
in SHADER
{
---
I noticed this problem after adding AoA support [1] to Ian's random UBO test
script [2].
[1] http://patchwork.freedesktop.org/patch/59956/
[2] http://cgit.freedesktop.org/~idr/piglit/log/?h=ubo-lolz
src/glsl/glsl_types.cpp | 13 +++--
1 file changed, 7 insertions(+), 6
https://bugs.freedesktop.org/show_bug.cgi?id=92054
Marcin Slusarz changed:
What|Removed |Added
Status|NEW |RESOLVED
Math operations in SandyBridge do not support source swizzling
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=92033
---
src/mesa/drivers/dri/i965/brw_ir_vec4.h | 3 ++-
src/mesa/drivers/dri/i965/brw_vec4.cpp | 11 +--
2 files changed, 11 insertions(+), 3 deletions(-)
diff
Thanks for reverting this.
The extension is required for importing dmabuf as textures in desktop
GL, so it can't removed no matter what the spec says.
Marek
On Wed, Sep 16, 2015 at 5:00 PM, Dave Airlie wrote:
> This reverts commit 48961fa3ba37999a6f8fd812458b735e39604a95.
>
Reviewed-by: Tapani Pälli
On 09/20/2015 01:50 AM, Jordan Justen wrote:
Move API validation to _mesa_validate_DispatchCompute in
api_validate.c.
Signed-off-by: Jordan Justen
---
src/mesa/main/api_validate.c | 44
On Fri, 2015-09-18 at 09:09 -0700, Kenneth Graunke wrote:
> On Friday, September 18, 2015 10:08:52 AM Iago Toral Quiroga wrote:
> > Until now we only used MRFs 1..15 for regular SEND messages, so the
> > message length could not possibly exceed the maximum size. Now that
> > we allow to use MRF
MOV instructions, as long as they don't have source modifiers, are
just copying bits around. So those kind of instruction could be
propagated even if there are type mismatches. This is needed because
NIR generates integer MOV instructions whenever it doesn't know what
else to generate.
This
https://bugs.freedesktop.org/show_bug.cgi?id=77449
Kai changed:
What|Removed |Added
Depends on||92059
--
You are
On Sun, Sep 20, 2015 at 8:48 AM, Antia Puentes wrote:
> Math operations in SandyBridge do not support source swizzling
I can't find any documentation to support this claim, but I remember
that SNB math must be in align1 mode so it can't do swizzles or
writemasking (see
GL_ARB_separate_shader_objects allow to match by name variable or block
interface. Input varying can't be removed as it is will impact the location
assignment.
The patch only disable deadcode optimization for inputs that don't have
an explicit location. It allows to remove various builtin
Current checks rely on the optimization phase to remove deadcode
varying. Therefore remaining varying are used.
Deadcode Optimizations won't be possible anymore with the support of
GL_ARB_separate_shader_objects. So it requires to check the used flag of
the varying.
Signed-off-by: Gregory
GL_ARB_separate_shader_objects requires that all inputs varying remains active.
This flag allow to distinguish input attribute from the input varying.
Flag will be used on next commit.
Signed-off-by: Gregory Hainaut
---
src/glsl/ast_to_hir.cpp | 4
On Sep 20, 2015 11:11 AM, "Matt Turner" wrote:
>
> On Sun, Sep 20, 2015 at 8:48 AM, Antia Puentes
wrote:
> > Math operations in SandyBridge do not support source swizzling
>
> I can't find any documentation to support this claim, but I remember
> that SNB
23 matches
Mail list logo