From: Dave Airlie
Noticed would fail, we were doing two things wrong
a) 1d arrays require the layers in height
b) minifying the layers field.
Signed-off-by: Dave Airlie
---
src/gallium/auxiliary/gallivm/lp_bld_sample_soa.c | 22 +-
src/gallium/drivers/llvmpipe/lp_setup.c
From: Dave Airlie
Now that we've gone over 7.
Signed-off-by: Dave Airlie
---
src/gallium/auxiliary/gallivm/lp_bld_sample.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/auxiliary/gallivm/lp_bld_sample.h
b/src/gallium/auxiliary/gallivm/lp_bld_sample.h
index dd
On 12/06/2012 02:57 PM, Ian Romanick wrote:
From: Ian Romanick
I erroneously added this back in January 2011 in commit 88421589.
Looking at the commit message, I have no idea why I added it. It only
added non-array structure fields to the symbol table, so array structure
fields are treated cor
On 12/07/2012 02:27 PM, Matt Turner wrote:
Otherwise messages say silly things like
glGetActiveUniformBlockiv(block index -1 >= 0)
---
src/mesa/main/uniforms.c |8
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/mesa/main/uniforms.c b/src/mesa/main/uniforms.
On 12/07/2012 04:34 PM, Matt Turner wrote:
The GL 3.1 and ES 3.0 specs say of glGetActiveUniformsiv:
"If an error occurs, nothing will be written to params."
So, make a pass through the indices and check that they're valid before
the pass that actually writes to params. Checking pname happen
On Fri, Dec 7, 2012 at 2:58 PM, Eric Anholt wrote:
> The limited performance testing I've done on this hasn't shown any
> statistically significant differences yet.
> ---
> .../dri/i965/brw_fs_schedule_instructions.cpp | 150
> +++-
> 1 file changed, 147 insertions(+), 3 de
The GL 3.1 and ES 3.0 specs say of glGetActiveUniformsiv:
"If an error occurs, nothing will be written to params."
So, make a pass through the indices and check that they're valid before
the pass that actually writes to params. Checking pname happens on the
first iteration of the second loop.
On Fri, Dec 7, 2012 at 4:09 PM, Brian Paul wrote:
>> The
>> only complication is that you have to check that the pnames are valid
>> too.
>
>
> Are you sure? On the first loop iteration we'll detect if pname is invalid
> and return before anything is written to params.
Oh, you are correct. I'd t
On 12/07/2012 04:59 PM, Matt Turner wrote:
On Fri, Dec 7, 2012 at 3:49 PM, Brian Paul wrote:
On 12/07/2012 04:23 PM, Matt Turner wrote:
The GL 3.1 and ES 3.0 specs say of glGetActiveUniformsiv:
"If an error occurs, nothing will be written to params."
So save a copy of params and restore
The GL 3.1 and ES 3.0 specs say of glGetActiveUniformsiv:
"If an error occurs, nothing will be written to params."
So, make a pass through the array and check for errors before the pass
that actually writes to params.
Fixes es3conform's getactiveuniformsiv_for_nonexistent_uniform_indices
test.
On Fri, Dec 7, 2012 at 3:49 PM, Brian Paul wrote:
> On 12/07/2012 04:23 PM, Matt Turner wrote:
>>
>> The GL 3.1 and ES 3.0 specs say of glGetActiveUniformsiv:
>> "If an error occurs, nothing will be written to params."
>>
>> So save a copy of params and restore it on error. The alternative is
Eric,
What do you think? Would set implemented via hash_table be
significantly less efficient?
-Jordan
On Fri, Dec 7, 2012 at 3:40 PM, Brian Paul wrote:
> Hi Jordan,
>
> set.c and hash_table.c look a LOT alike. Could the functions in set.c be
> implemented in terms of the hash_table.c function
On 12/07/2012 04:23 PM, Matt Turner wrote:
The GL 3.1 and ES 3.0 specs say of glGetActiveUniformsiv:
"If an error occurs, nothing will be written to params."
So save a copy of params and restore it on error. The alternative is
making a pass through the array to check for errors before a seco
Hi Jordan,
set.c and hash_table.c look a LOT alike. Could the functions in set.c
be implemented in terms of the hash_table.c functions? It seems
little silly to have so much duplicated code.
-Brian
___
mesa-dev mailing list
mesa-dev@lists.freedesk
The GL 3.1 and ES 3.0 specs say of glGetActiveUniformsiv:
"If an error occurs, nothing will be written to params."
So save a copy of params and restore it on error. The alternative is
making a pass through the array to check for errors before a second pass
that actually writes *params, but that
On 12/07/2012 03:26 PM, Matt Turner wrote:
On Fri, Dec 7, 2012 at 2:11 PM, Brian Paul wrote:
On 12/07/2012 10:40 AM, Chad Versace wrote:
On 12/07/2012 01:29 PM, dmi...@freedesktop.org wrote:
./configure --enable-debug seems to leave -O2 omptimization flag
turned on.
Is there a quick method
On 12/07/2012 03:27 PM, Matt Turner wrote:
Otherwise messages say silly things like
glGetActiveUniformBlockiv(block index -1>= 0)
---
src/mesa/main/uniforms.c |8
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/mesa/main/uniforms.c b/src/mesa/main/uniforms.c
I've tested this to be true with various ALU ops on gen7 (with the
exception of MADs, which go at either 3 or 4 cycles per dispatch).
---
.../dri/i965/brw_fs_schedule_instructions.cpp| 18 +++---
1 file changed, 15 insertions(+), 3 deletions(-)
diff --git a/src/mesa/drivers/
The limited performance testing I've done on this hasn't shown any
statistically significant differences yet.
---
.../dri/i965/brw_fs_schedule_instructions.cpp | 150 +++-
1 file changed, 147 insertions(+), 3 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/brw_fs_schedul
For gen7 everything changes, and we have actual information on latency.
---
.../dri/i965/brw_fs_schedule_instructions.cpp | 74 +++-
1 file changed, 41 insertions(+), 33 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/brw_fs_schedule_instructions.cpp
b/src/mesa/drivers
This gives the instruction scheduler a chance to schedule between the
loads, whereas before it was restricted due to the dependencies between
the MRFs for setting them up.
For one shader in gles3conform, it goes from getting stuck in register
allocation for as long as anybody's bothered to leave i
---
src/mesa/drivers/dri/i965/brw_fs.cpp |4 +-
src/mesa/drivers/dri/i965/brw_fs.h |2 +-
.../dri/i965/brw_fs_schedule_instructions.cpp | 90
3 files changed, 78 insertions(+), 18 deletions(-)
diff --git a/src/mesa/drivers/dri/i965
This came from an idea by Ben Segovia. 16-wide pixel shaders are very
important for latency hiding on i965, so we want to try really hard to
get them. If scheduling an instruction makes some set of instructions
available, those are probably the ones that make the instruction's
result dead. By ch
Seeing when instructions become available to schedule is really useful.
---
.../dri/i965/brw_fs_schedule_instructions.cpp | 21
1 file changed, 21 insertions(+)
diff --git a/src/mesa/drivers/dri/i965/brw_fs_schedule_instructions.cpp
b/src/mesa/drivers/dri/i965/brw_fs_
Here's the series that came out of trying to fix the gles3conform ubo test
that has a steaming pile of ubo variables and was dying in register
allocation. Patch 4/7 is where that got fixed.
Patches 5-7 are followons that I've wanted and had in various forms lying
around my trees, only now with no
On Fri, Dec 7, 2012 at 2:11 PM, Brian Paul wrote:
> On 12/07/2012 10:40 AM, Chad Versace wrote:
>>
>> On 12/07/2012 01:29 PM, dmi...@freedesktop.org wrote:
>>>
>>> ./configure --enable-debug seems to leave -O2 omptimization flag
>>> turned on.
>>>
>>> Is there a quick method to switch it off?
>>
>
From: Tom Stellard
---
lib/Target/AMDGPU/AMDILISelDAGToDAG.cpp| 5 +++
.../AMDGPU/MCTargetDesc/R600MCCodeEmitter.cpp | 2 ++
lib/Target/AMDGPU/R600ISelLowering.cpp | 25 +++---
lib/Target/AMDGPU/R600Instructions.td | 38 --
From: Tom Stellard
---
lib/Target/AMDGPU/InstPrinter/AMDGPUInstPrinter.cpp | 2 ++
lib/Target/AMDGPU/R600Instructions.td | 14 +-
test/CodeGen/R600/load.constant_addrspace.f32.ll| 2 +-
test/CodeGen/R600/load.i8.ll| 2 +-
4 files changed,
From: Tom Stellard
This will reduce the number of tablegen patterns we need.
---
lib/Target/AMDGPU/AMDGPUISelLowering.cpp | 14 +
.../AMDGPU/MCTargetDesc/R600MCCodeEmitter.cpp | 11 ++--
lib/Target/AMDGPU/R600ISelLowering.cpp | 2 -
lib/Target/AMDGPU/R600Instructi
From: Tom Stellard
---
lib/CodeGen/SelectionDAG/LegalizeDAG.cpp | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git a/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
b/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
index 9946694..2596f00 100644
--- a/lib/CodeGen/SelectionDAG/LegalizeDAG.c
From: Tom Stellard
---
lib/CodeGen/SelectionDAG/LegalizeDAG.cpp | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
b/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp
index abf40b7..9946694 100644
--- a/lib/CodeGen/SelectionDAG/LegalizeDAG
Otherwise messages say silly things like
glGetActiveUniformBlockiv(block index -1 >= 0)
---
src/mesa/main/uniforms.c |8
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/mesa/main/uniforms.c b/src/mesa/main/uniforms.c
index 3f156bf..77b195e 100644
--- a/src/mesa/m
On 12/07/2012 10:40 AM, Chad Versace wrote:
On 12/07/2012 01:29 PM, dmi...@freedesktop.org wrote:
./configure --enable-debug seems to leave -O2 omptimization flag
turned on.
Is there a quick method to switch it off?
During the ./configure&& make&& make install process, I set
export CFLAGS=
I had tried this in the past, but ran into trouble with applications that
sample from undiscarded pixels in the same subspan. To fix that issue, only
jump to the end for an entire subspan at a time.
Improves GLbenchmark 2.7 (1024x768) performance by 7.9 +/- 1.5% (n=8).
---
src/mesa/drivers/dri/i
This makes much more sense on gen6+, and will also prove useful for
early exit of shaders on discard.
---
src/mesa/drivers/dri/i965/brw_defines.h|1 -
src/mesa/drivers/dri/i965/brw_fs.cpp |8 +++
src/mesa/drivers/dri/i965/brw_fs.h |1 -
src/mes
We're going to redo discard handling to track discards in the other flag
subregister, saving instructions in the discard and allowing predicated
jumps out to the end of the shader.
---
src/mesa/drivers/dri/i965/brw_eu.c |6 ++
src/mesa/drivers/dri/i965/brw_eu.h
We've been calling it a register number, it's actually the subregister,
and things will get confusing once we start using it if it isn't fixed.
---
src/mesa/drivers/dri/i965/brw_disasm.c |4 ++--
src/mesa/drivers/dri/i965/brw_eu_compact.c |4 ++--
src/mesa/drivers/dri/i965/brw_struct
We're about to start using the f0.1 subregister.
---
src/mesa/drivers/dri/i965/brw_blorp_blit.cpp |2 +-
src/mesa/drivers/dri/i965/brw_eu.c |2 +-
src/mesa/drivers/dri/i965/brw_eu.h |6 +++---
src/mesa/drivers/dri/i965/brw_fs_emit.cpp|4 ++--
4 files changed
This makes our output more consistent with other disasm tools, and
will be necessary when we start using f0.1.
---
src/mesa/drivers/dri/i965/brw_disasm.c | 16 +++-
1 file changed, 15 insertions(+), 1 deletion(-)
diff --git a/src/mesa/drivers/dri/i965/brw_disasm.c
b/src/mesa/driver
---
src/mesa/drivers/dri/i965/brw_disasm.c |2 +-
src/mesa/drivers/dri/i965/brw_structs.h | 12
2 files changed, 9 insertions(+), 5 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/brw_disasm.c
b/src/mesa/drivers/dri/i965/brw_disasm.c
index 1f3aeca..b18676d 100644
--- a/sr
There's a flag subreg nr field in bits2 next to src0.vertstride, but
there shouldn't be anything in bits3 next to src1.vertstride.
---
src/mesa/drivers/dri/i965/brw_structs.h |6 ++
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/brw_structs.h
b/src
This patch does two things:
1. Constant buffer state changes were broken (but happened to work by
dumb luck). The problem is we weren't calling draw_do_flush() in
draw_set_mapped_constant_buffer() when we changed that state. All the
other draw_set_foo() functions were calling draw_do_fl
---
src/gallium/auxiliary/draw/draw_pt.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/src/gallium/auxiliary/draw/draw_pt.c
b/src/gallium/auxiliary/draw/draw_pt.c
index 23cdf36..7113b9e 100644
--- a/src/gallium/auxiliary/draw/draw_pt.c
+++ b/src/gallium/auxiliary/draw/
---
src/gallium/auxiliary/draw/draw_private.h |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/src/gallium/auxiliary/draw/draw_private.h
b/src/gallium/auxiliary/draw/draw_private.h
index 86ce397..e52b3fd 100644
--- a/src/gallium/auxiliary/draw/draw_private.h
+++ b/src/gal
---
src/gallium/auxiliary/draw/draw_pt.c |5 ++---
1 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/src/gallium/auxiliary/draw/draw_pt.c
b/src/gallium/auxiliary/draw/draw_pt.c
index 08e06e8..23cdf36 100644
--- a/src/gallium/auxiliary/draw/draw_pt.c
+++ b/src/gallium/auxiliary/dr
To clean it up and make it look more like the non-LLVM
fetch_pipeline_prepare() function.
---
.../draw/draw_pt_fetch_shade_pipeline_llvm.c | 140 ++-
1 files changed, 74 insertions(+), 66 deletions(-)
diff --git a/src/gallium/auxiliary/draw/draw_pt_fetch_shade_pipeline_llvm
When one function is changed, also look at the other.
Presently, there are some differences with respect to geometry
shaders and instanced drawing...
---
.../auxiliary/draw/draw_pt_fetch_shade_pipeline.c |6 ++
.../draw/draw_pt_fetch_shade_pipeline_llvm.c |5 +
2 files chang
Ian Romanick writes:
> From: Paul Berry
>
> ---
> src/mesa/main/ff_fragment_shader.cpp | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/src/mesa/main/ff_fragment_shader.cpp
> b/src/mesa/main/ff_fragment_shader.cpp
> index 3a80ab7..02b4707 100644
> --- a/src/mesa/main/ff_fragment_shader
The http://cgit.freedesktop.org/~airlied/mesa/log/?h=gallium-tbo branch looks
good to me.
There is some work necessary to support TXF from texture buffers in softpipe,
but this definitely seems like a step forward.
Jose
- Original Message -
> Hi,
>
> So I got UBOs working with softpip
>
> During the ./configure && make && make install process, I set
>
> export CFLAGS="-O0 -g3"
> export CXXFLAGS="-O0 -g3"
>
>
It works. Thanks!
--
With best regards,
Dmitry
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedeskt
On 12/07/2012 01:29 PM, dmi...@freedesktop.org wrote:
> ./configure --enable-debug seems to leave -O2 omptimization flag
> turned on.
>
> Is there a quick method to switch it off?
During the ./configure && make && make install process, I set
export CFLAGS="-O0 -g3"
export CXXFLAGS="-O0 -g3"
The
Pushed, thanks!
Marek
On Fri, Dec 7, 2012 at 5:38 PM, Stefan Dösinger wrote:
> The read can remain disabled if the src alpha factor needs it because
> the result would still be zero.
>
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57984
>
> NOTE: This is a candidate for stable release
The read can remain disabled if the src alpha factor needs it because
the result would still be zero.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=57984
NOTE: This is a candidate for stable release branches.
---
src/gallium/drivers/r300/r300_state.c | 12 ++--
1 files changed
./configure --enable-debug seems to leave -O2 omptimization flag
turned on.
Is there a quick method to switch it off?
Thanks.
--
With best regards,
Dmitry
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/li
Thanks Michel!
Reviewed-by: Jose Fonseca
- Original Message -
> From: Michel Dänzer
>
> Signed-off-by: Michel Dänzer
> ---
> src/gallium/targets/egl-static/Makefile |4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/src/gallium/targets/egl-static/Makefi
From: Michel Dänzer
Signed-off-by: Michel Dänzer
---
src/gallium/targets/egl-static/Makefile |4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/gallium/targets/egl-static/Makefile
b/src/gallium/targets/egl-static/Makefile
index 08626a9..0f41a21 100644
--- a/src/gall
56 matches
Mail list logo