On 05/08/2015 09:56 AM, Pohjolainen, Topi wrote:
On Fri, May 08, 2015 at 09:51:54AM +0300, Tapani P?lli wrote:
This fixes bugs with special cases where we have arrays of
structures containing samplers or arrays of samplers.
I've verified that patch results in calculating same index value as
r
>
>
> On 05/08/2015 12:13 AM, Ian Romanick wrote:
>> On 05/07/2015 12:57 AM, Marta Lofstedt wrote:
>>> From: Marta Lofstedt
>>>
>>> v2: only expose enums from GL_ARB_shader_image_load_store
>>> for gles 3.1 and GL core
>>>
>>> Signed-off-by: Marta Lofstedt
>>> ---
>>> src/mesa/main/get.c
On Thursday, May 07, 2015 06:17:46 PM Matt Turner wrote:
> On Thu, May 7, 2015 at 4:50 PM, Jason Ekstrand wrote:
> > GLSL IR vs. NIR shader-db results for SIMD8 vertex shaders on Broadwell:
> >
> >total instructions in shared programs: 2724483 -> 2711790 (-0.47%)
> >instructions in affecte
That is strange, it was introduced in fb370f89d but then has gone missing ..
Reviewed-by: Tapani Pälli
On 05/07/2015 06:13 PM, Marta Lofstedt wrote:
The return type for GL_VERTEX_BINDING_STRIDE is missing,
this cause glGetIntegeri_v to fail.
Signed-off-by: Marta Lofstedt
---
src/mesa/main
According to Glenn, shifts on R600 have 5x the throughput as multiplies.
Intel GPUs have strange integer multiplication restrictions - on most
hardware, MUL actually only does a 32-bit x 16-bit multiply. This
means the arguments aren't commutative, which can limit our constant
propagation options
On 05/07/2015 05:46 PM, Pohjolainen, Topi wrote:
> On Thu, May 07, 2015 at 04:43:21PM +0300, Pohjolainen, Topi wrote:
>> On Tue, Apr 28, 2015 at 11:08:00PM +0300, Abdiel Janulgue wrote:
>>> This patch implements the binding table enable command which is also
>>> used to allocate a binding table p
On 05/07/2015 05:48 PM, Pohjolainen, Topi wrote:
> On Tue, Apr 28, 2015 at 11:08:23PM +0300, Abdiel Janulgue wrote:
>> Programming null constants with gather constant tables seems to
>> be unsupported and results in a GPU lockup even with the prescribed
>> GPU workarounds in the bspec. Found out
On 05/07/2015 04:43 PM, Pohjolainen, Topi wrote:
> On Tue, Apr 28, 2015 at 11:08:00PM +0300, Abdiel Janulgue wrote:
>> This patch implements the binding table enable command which is also
>> used to allocate a binding table pool where hardware-generated
>> binding table entries are flushed into.
perjantai 8. toukokuuta 2015 Ian Romanick kirjoitti:
> On 05/07/2015 05:21 AM, Pohjolainen, Topi wrote:
> > On Tue, May 05, 2015 at 02:25:29PM +0300, Juha-Pekka Heikkila wrote:
> >> Stop context creation if something failed. If something errored
> >> during context creation we'd segfault. Now wil
On 05/07/2015 06:17 PM, Pohjolainen, Topi wrote:
> On Tue, Apr 28, 2015 at 11:08:08PM +0300, Abdiel Janulgue wrote:
>> The resource streamer is able to gather and pack sparsely-located
>> constant data from any buffer object by a referring to a gather table
>> This patch adds support for keeping
On Fri, May 08, 2015 at 03:23:52PM +0300, Juha-Pekka Heikkil? wrote:
>perjantai 8. toukokuuta 2015 Ian Romanick kirjoitti:
>
> On 05/07/2015 05:21 AM, Pohjolainen, Topi wrote:
> > On Tue, May 05, 2015 at 02:25:29PM +0300, Juha-Pekka Heikkila wrote:
> >> Stop context creation if
Le 06/05/2015 03:00, Dave Airlie a écrit :
On 2 May 2015 at 20:15, Axel Davy wrote:
Only EGL_WINDOW_BIT is supported. Remove tests related.
Is this there no plans to support pixmap/pbuffer/ or any of the other bits?
Seems like a step in the wrong direction if we really should be supporting
ot
Connor Abbott writes:
> On Tue, May 5, 2015 at 4:29 PM, Francisco Jerez wrote:
>> ---
>> src/glsl/nir/glsl_to_nir.cpp | 125
>> +++
>> 1 file changed, 114 insertions(+), 11 deletions(-)
>>
>> diff --git a/src/glsl/nir/glsl_to_nir.cpp b/src/glsl/nir/glsl_
Connor Abbott writes:
> On IRC, Ken and I were discussing using a scheme inspired by SPIR-V,
> which has an OpImagePointer instruction that forms a pointer to the
> particular texel of the image as well as
> OpAtomic{Load,Store,Exchange,etc.} that operate on an image or shared
> buffer pointer. T
Hello all,
On 21 April 2015 at 13:40, Emil Velikov wrote:
> Hi all,
>
> Here is the preliminary release schedule for Mesa 10.6. Personally I
> would love if we can go up to Mesa 11.0 (Spinal Tap anyone ?) although
> it might happen with the September release.
>
> May 15th 2015 - Feature freeze/Re
On 2 May 2015 at 11:15, Axel Davy wrote:
> the wl_registry and the wl_queue allocated weren't destroyed.
>
Would this have any affect other than leaking memory ? If so can you
please add the mesa-stable tag (prior to pushing).
Thanks
Emil
___
mesa-dev m
On 2 May 2015 at 11:15, Axel Davy wrote:
> Checks blitImage is implemented.
> Initially having the __DRIimageExtension extension
> at version 9 at least meant blitImage was supported.
> However some implementations do advertise version >= 9
> without implementing it.
>
Afaict in the above mentione
Previously, this case was being handled in match_expression prior to
calling match_value. However, there is really no good reason for this
given that match_value has all of the information it needs. Also, they
weren't being handled properly in the commutative case and putting it in
match_value gi
On 2 May 2015 at 11:15, Axel Davy wrote:
> It is possible the server advertises a render-node.
> In that case no authentication is needed,
> and Gem names are forbidden.
>
> Signed-off-by: Axel Davy
> ---
> diff --git a/src/egl/drivers/dri2/platform_wayland.c
> b/src/egl/drivers/dri2/platform_w
On Fri, May 8, 2015 at 10:30 AM, Francisco Jerez wrote:
> Connor Abbott writes:
>
>> On Tue, May 5, 2015 at 4:29 PM, Francisco Jerez
>> wrote:
>>> ---
>>> src/glsl/nir/glsl_to_nir.cpp | 125
>>> +++
>>> 1 file changed, 114 insertions(+), 11 deletions(-)
v2: Undefine coordinate components not applicable to the target.
---
src/glsl/nir/glsl_to_nir.cpp | 126 +++
1 file changed, 115 insertions(+), 11 deletions(-)
diff --git a/src/glsl/nir/glsl_to_nir.cpp b/src/glsl/nir/glsl_to_nir.cpp
index f6b8331..fab4508
Makes sense.
Reviewed-by: Connor Abbott
On Fri, May 8, 2015 at 11:38 AM, Jason Ekstrand wrote:
> Previously, this case was being handled in match_expression prior to
> calling match_value. However, there is really no good reason for this
> given that match_value has all of the information it n
Update the comment in patch 1, and the entire thing is
Reviewed-by: Connor Abbott
On Fri, May 8, 2015 at 12:40 PM, Francisco Jerez wrote:
> v2: Undefine coordinate components not applicable to the target.
>
> ---
> src/glsl/nir/glsl_to_nir.cpp | 126
> +++--
On 2 May 2015 at 11:15, Axel Davy wrote:
> When the server gpu and requested gpu are different:
> . They likely don't support the same tiling modes
> . They likely do not have fast access to the same locations
>
> Thus we do:
> . render to a tiled buffer we do not share with the server
> . Copy th
Shader-db results for fragment shaders on Broadwell:
total instructions in shared programs: 4310987 -> 4310663 (-0.01%)
instructions in affected programs: 43242 -> 42918 (-0.75%)
helped:142
HURT: 0
Shader-db results
On Fri, May 8, 2015 at 3:36 AM, Kenneth Graunke wrote:
> According to Glenn, shifts on R600 have 5x the throughput as multiplies.
>
> Intel GPUs have strange integer multiplication restrictions - on most
> hardware, MUL actually only does a 32-bit x 16-bit multiply. This
> means the arguments are
On Fri, May 8, 2015 at 3:27 AM, Kenneth Graunke wrote:
> On Thursday, May 07, 2015 06:17:46 PM Matt Turner wrote:
>> On Thu, May 7, 2015 at 4:50 PM, Jason Ekstrand wrote:
>> > GLSL IR vs. NIR shader-db results for SIMD8 vertex shaders on Broadwell:
>> >
>> >total instructions in shared progra
The opt_sampler_eot optimisation seems to break when the last
instruction is SHADER_OPCODE_TG4. A bunch of Piglit tests end up doing
this so it causes a lot of regressions. I can't find any documentation
or known workarounds to indicate that this is expected behaviour, but
considering that this is
On Fri, May 8, 2015 at 10:04 AM, Jason Ekstrand wrote:
> Shader-db results for fragment shaders on Broadwell:
>
>total instructions in shared programs: 4310987 -> 4310663 (-0.01%)
>instructions in affected programs: 43242 -> 42918 (-0.75%)
>helped:14
On Fri, May 8, 2015 at 10:04 AM, Jason Ekstrand wrote:
> Shader-db results for fragment shaders on Broadwell:
>
>total instructions in shared programs: 4310987 -> 4310663 (-0.01%)
>instructions in affected programs: 43242 -> 42918 (-0.75%)
>helped:14
On Fri, May 8, 2015 at 10:08 AM, Jason Ekstrand wrote:
> On Fri, May 8, 2015 at 3:27 AM, Kenneth Graunke wrote:
>> On Thursday, May 07, 2015 06:17:46 PM Matt Turner wrote:
>>> On Thu, May 7, 2015 at 4:50 PM, Jason Ekstrand wrote:
>>> > GLSL IR vs. NIR shader-db results for SIMD8 vertex shaders o
On 2 May 2015 at 11:15, Axel Davy wrote:
> Enables to use dri config for swrast, like vblank_mode.
>
> Signed-off-by: Axel Davy
> ---
> src/egl/drivers/dri2/egl_dri2.c| 21 ++---
> src/gallium/state_trackers/dri/drisw.c | 1 +
> src/mesa/drivers/dri/swrast/swrast.c |
On 8 May 2015 at 18:06, Axel Davy wrote:
> On Fri, 8 May 2015, Emil Velikov wrote:
>
>> Shouldn't we authenticate with the correct gpu or master/render node ?
>> This implementation will auth with GPU1, and then use GPU2 which seems
>> a bit odd. I might be missing something ?
>>
>>
>
> The origin
On 8 May 2015 at 14:34, Axel Davy wrote:
> Le 06/05/2015 03:00, Dave Airlie a écrit :
>>
>> On 2 May 2015 at 20:15, Axel Davy wrote:
>>>
>>> Only EGL_WINDOW_BIT is supported. Remove tests related.
>>
>> Is this there no plans to support pixmap/pbuffer/ or any of the other
>> bits?
>>
>> Seems lik
Tom Stellard writes:
> Events can be added to an OpenCL command queue concurrently from
> multiple threads, but pipe_context and pipe_screen objects
> are not threadsafe. The threadsafe wrappers protect all pipe_screen
> and pipe_context function calls with a mutex, so we can safely use
> them w
On 8 May 2015 at 11:36, Tapani Pälli wrote:
> That is strange, it was introduced in fb370f89d but then has gone missing ..
>
Seems like it broke shortly after it was introduced (around 10.1.0-devel)
commit 902f9df36bec7d67a2d8bc4c24d89d9d57964903
Author: Francisco Jerez
Date: Mon Nov 25 10:11:
On 8 May 2015 at 03:10, Michel Dänzer wrote:
> On 08.05.2015 03:24, Tom Stellard wrote:
>> For this particular situation, I'm happy with any solution that:
>>
>> 1. Allows a user to install the icd file to /etc if he or she wants to.
>
> --sysconfdir=/etc
>
> That covers drirc as well.
>
>
>> 2. D
I thought it might be a good idea to try posting these patches again
since it's been 6 months since they were originally posted. The
patches are a lot more useful now since the command parser in the
kernel is working correctly for Haswell. This means the functionality
is no longer restricted to onl
Previously whenever a primitive is drawn the driver would call
_mesa_check_conditional_render which blocks waiting for the result of
the query to determine whether to render. On Gen7+ there is a bit in
the 3DPRIMITIVE command which can be used to disable the primitive
based on the value of a state
On Friday, May 08, 2015 10:18:44 AM Matt Turner wrote:
> On Fri, May 8, 2015 at 10:04 AM, Jason Ekstrand wrote:
> > Shader-db results for fragment shaders on Broadwell:
> >
> >total instructions in shared programs: 4310987 -> 4310663 (-0.01%)
> >instructions in affected programs: 43242
In order to detect whether the predicate source registers can be used
in a later patch we will need to know the version number for the
command parser. This patch just adds a member to intel_screen and does
an ioctl to get the version.
Reviewed-by: Kenneth Graunke
---
src/mesa/drivers/dri/i965/in
On 05/08/2015 03:36 AM, Kenneth Graunke wrote:
> According to Glenn, shifts on R600 have 5x the throughput as multiplies.
>
> Intel GPUs have strange integer multiplication restrictions - on most
> hardware, MUL actually only does a 32-bit x 16-bit multiply. This
> means the arguments aren't comm
On Fri, May 8, 2015 at 11:05 AM, Kenneth Graunke wrote:
> On Friday, May 08, 2015 10:18:44 AM Matt Turner wrote:
>> On Fri, May 8, 2015 at 10:04 AM, Jason Ekstrand wrote:
>> > Shader-db results for fragment shaders on Broadwell:
>> >
>> >total instructions in shared programs: 4310987 -> 43106
On Fri, May 8, 2015 at 11:05 AM, Kenneth Graunke wrote:
> On Friday, May 08, 2015 10:18:44 AM Matt Turner wrote:
>> On Fri, May 8, 2015 at 10:04 AM, Jason Ekstrand wrote:
>> > Shader-db results for fragment shaders on Broadwell:
>> >
>> >total instructions in shared programs: 4310987 -> 43106
On Tuesday, May 05, 2015 11:29:52 PM Francisco Jerez wrote:
> ---
> src/glsl/nir/nir_intrinsics.h | 27 +++
> 1 file changed, 27 insertions(+)
>
> diff --git a/src/glsl/nir/nir_intrinsics.h b/src/glsl/nir/nir_intrinsics.h
> index 8e28765..4b13c75 100644
> --- a/src/glsl/ni
On Fri, May 8, 2015 at 11:13 AM, Jason Ekstrand wrote:
> On Fri, May 8, 2015 at 11:05 AM, Kenneth Graunke
> wrote:
>> On Friday, May 08, 2015 10:18:44 AM Matt Turner wrote:
>>> On Fri, May 8, 2015 at 10:04 AM, Jason Ekstrand
>>> wrote:
>>> > Shader-db results for fragment shaders on Broadwell:
On Fri, May 8, 2015 at 11:15 AM, Matt Turner wrote:
> On Fri, May 8, 2015 at 11:13 AM, Jason Ekstrand wrote:
>> On Fri, May 8, 2015 at 11:05 AM, Kenneth Graunke
>> wrote:
>>> On Friday, May 08, 2015 10:18:44 AM Matt Turner wrote:
On Fri, May 8, 2015 at 10:04 AM, Jason Ekstrand
wrote
On Fri, May 8, 2015 at 6:36 AM, Kenneth Graunke wrote:
> + # Multiplication by 4 comes up fairly often in indirect offset
> calculations.
> + # Some GPUs have weird integer multiplication limitations, but shifts
> should work
> + # equally well everywhere.
> + (('imul', 4, a), ('ishl', a
Hi Tom,
On 8 May 2015 at 00:36, Tom Stellard wrote:
> Events can be added to an OpenCL command queue concurrently from
> multiple threads, but pipe_context and pipe_screen objects
> are not threadsafe. The threadsafe wrappers protect all pipe_screen
> and pipe_context function calls with a mutex
On Friday, May 08, 2015 07:04:31 PM Neil Roberts wrote:
> Previously whenever a primitive is drawn the driver would call
> _mesa_check_conditional_render which blocks waiting for the result of
> the query to determine whether to render. On Gen7+ there is a bit in
> the 3DPRIMITIVE command which can
I don't have CHV or SKL hw or docs to try and confirm this, but this
does what it claims to.
Reviewed-by: Chris Forbes
On Sat, May 9, 2015 at 5:10 AM, Neil Roberts wrote:
> The opt_sampler_eot optimisation seems to break when the last
> instruction is SHADER_OPCODE_TG4. A bunch of Piglit tests
total instructions in shared programs: 7152330 -> 7137006 (-0.21%)
instructions in affected programs: 1330548 -> 1315224 (-1.15%)
helped:5797
HURT: 76
GAINED:0
LOST:
On Fri, May 8, 2015 at 11:53 AM, Jason Ekstrand wrote:
> total instructions in shared programs: 7152330 -> 7137006 (-0.21%)
> instructions in affected programs: 1330548 -> 1315224 (-1.15%)
> helped:5797
> HURT: 76
I'm doing some
On Fri, May 8, 2015 at 11:11 AM, Ian Romanick wrote:
> On 05/08/2015 03:36 AM, Kenneth Graunke wrote:
>> According to Glenn, shifts on R600 have 5x the throughput as multiplies.
>>
>> Intel GPUs have strange integer multiplication restrictions - on most
>> hardware, MUL actually only does a 32-bit
---
src/glsl/nir/nir_search.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/glsl/nir/nir_search.c b/src/glsl/nir/nir_search.c
index 5ba0160..d86655b 100644
--- a/src/glsl/nir/nir_search.c
+++ b/src/glsl/nir/nir_search.c
@@ -90,6 +90,7 @@ match_value(const nir_search_value *value, nir_alu
Ilia Mirkin writes:
> On Fri, May 8, 2015 at 6:36 AM, Kenneth Graunke wrote:
>> + # Multiplication by 4 comes up fairly often in indirect offset
>> calculations.
>> + # Some GPUs have weird integer multiplication limitations, but shifts
>> should work
>> + # equally well everywhere.
>> +
Over-all, I think this is on the right track, but I still don't think
it's 100% correct.
On Fri, May 8, 2015 at 12:04 AM, Tapani Pälli wrote:
>
>
> On 05/08/2015 09:56 AM, Pohjolainen, Topi wrote:
>>
>> On Fri, May 08, 2015 at 09:51:54AM +0300, Tapani P?lli wrote:
>>>
>>> This fixes bugs with spe
Patches 11-13 are:
Reviewed-by: Chris Forbes
On Fri, Apr 24, 2015 at 1:42 PM, Dave Airlie wrote:
> From: Dave Airlie
>
> This type will be used to store the name of subroutine types
>
> as in subroutine void myfunc(void);
> will store myfunc into a subroutine type.
>
> This is required to the
Hi all,
Just had a quick look at Debian's repo and noticed something rather
worrying - the ABI of our libEGL is not stable.
Stable in the sense of - we provide a growing list of static symbols
for each new function an EGL extension adds.
This comes as we set EGL_EGLEXT_PROTOTYPES, prior to includ
On Fri, May 8, 2015 at 2:53 PM, Jason Ekstrand wrote:
> total instructions in shared programs: 7152330 -> 7137006 (-0.21%)
> instructions in affected programs: 1330548 -> 1315224 (-1.15%)
> helped:5797
> HURT: 76
> GAINED:
Reviewed-by: Connor Abbott
On Fri, May 8, 2015 at 3:13 PM, Jason Ekstrand wrote:
> ---
> src/glsl/nir/nir_search.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/src/glsl/nir/nir_search.c b/src/glsl/nir/nir_search.c
> index 5ba0160..d86655b 100644
> --- a/src/glsl/nir/nir_search.c
> +
Tom Stellard writes:
> This fixes a potential crash where on a sequence like this:
>
> Thread 0: Check if queue is not empty.
> Thread 1: Remove item from queue, making it empty.
> Thread 0: Do something assuming queue is not empty.
> ---
> src/gallium/state_trackers/clover/core/queue.cpp | 2 ++
On 05/08/2015 11:55 AM, Jason Ekstrand wrote:
> On Fri, May 8, 2015 at 11:53 AM, Jason Ekstrand wrote:
>> total instructions in shared programs: 7152330 -> 7137006 (-0.21%)
>> instructions in affected programs: 1330548 -> 1315224 (-1.15%)
>> helped:5797
>> HURT:
On Fri, May 8, 2015 at 3:20 PM, Ian Romanick wrote:
> On 05/08/2015 11:55 AM, Jason Ekstrand wrote:
>> On Fri, May 8, 2015 at 11:53 AM, Jason Ekstrand wrote:
>>> total instructions in shared programs: 7152330 -> 7137006 (-0.21%)
>>> instructions in affected programs: 1330548 -> 1315224 (-1.15
On 05/08/2015 03:25 PM, Jason Ekstrand wrote:
> On Fri, May 8, 2015 at 3:20 PM, Ian Romanick wrote:
>> On 05/08/2015 11:55 AM, Jason Ekstrand wrote:
>>> On Fri, May 8, 2015 at 11:53 AM, Jason Ekstrand
>>> wrote:
total instructions in shared programs: 7152330 -> 7137006 (-0.21%)
instruc
On 05/08/2015 03:31 PM, Ian Romanick wrote:
> On 05/08/2015 03:25 PM, Jason Ekstrand wrote:
>> On Fri, May 8, 2015 at 3:20 PM, Ian Romanick wrote:
>>> On 05/08/2015 11:55 AM, Jason Ekstrand wrote:
On Fri, May 8, 2015 at 11:53 AM, Jason Ekstrand
wrote:
> total instructions in shared
On Fri, May 8, 2015 at 3:32 PM, Ian Romanick wrote:
> On 05/08/2015 03:31 PM, Ian Romanick wrote:
>> On 05/08/2015 03:25 PM, Jason Ekstrand wrote:
>>> On Fri, May 8, 2015 at 3:20 PM, Ian Romanick wrote:
On 05/08/2015 11:55 AM, Jason Ekstrand wrote:
> On Fri, May 8, 2015 at 11:53 AM, Jaso
From: Ian Romanick
Shader-db results:
GM45 and Iron Lake:
total instructions in shared programs: 7888585 -> 7888585 (0.00%)
instructions in affected programs: 0 -> 0
Sandy Bridge, Ivy Bridge, Haswell, and Broadwell:
total instructions in shared programs: 9598608 -> 9598572 (-0.00%)
instruct
This was not fully baked. I'll send out a fixed version later.
On 05/08/2015 07:05 PM, Ian Romanick wrote:
> From: Ian Romanick
>
> Shader-db results:
>
> GM45 and Iron Lake:
> total instructions in shared programs: 7888585 -> 7888585 (0.00%)
> instructions in affected programs: 0 -> 0
>
On 01/19/2015 12:00 PM, Chris Wilson wrote:
For enable BufferAge support, we just have to be not using the
DRI2Buffer->flags field for any purpose (i.e. it is always expected to
be 0, as it is now) and to be sure to swap the flags field whenever we
exchange buffers. As nouveau does not exactly
I found this useful while debugging some control flow bugs while working
on the dead control flow pass.
Signed-off-by: Connor Abbott
---
src/glsl/nir/nir_validate.c | 21 +
1 file changed, 21 insertions(+)
diff --git a/src/glsl/nir/nir_validate.c b/src/glsl/nir/nir_validate.
We were already doing this internally for iterating over a function
implementation, so just expose it directly.
Signed-off-by: Connor Abbott
---
src/glsl/nir/nir.c | 7 +++
src/glsl/nir/nir.h | 2 ++
2 files changed, 9 insertions(+)
diff --git a/src/glsl/nir/nir.c b/src/glsl/nir/nir.c
index
This series implements a bunch of related optimizations that run at
once as part of the same pass to eliminate control flow that is
useless. This includes both unreachable code and useless loops, i.e.
loops that don't compute a result used by the rest of the code. To do
this, I needed to change som
We were rewriting the uses of the intrinsic instruction to point to
something else after removing it, which only happened to work since we
were lax about having dangling uses when removing instructions. Fix that
up.
Signed-off-by: Connor Abbott
---
src/glsl/nir/nir_lower_vars_to_ssa.c | 3 ++-
1
The point of cleanup_defs_uses() is to make an instruction safe to
remove by removing any references that the rest of the shader may have
to it. Previously, it was handling register use/def sets and removing
the instruction from the use sets of any SSA sources it had, but if the
instruction defined
Doesn't do anything on the public shader-db.
Signed-off-by: Connor Abbott
---
src/glsl/Makefile.sources | 1 +
src/mesa/drivers/dri/i965/brw_nir.c | 2 ++
2 files changed, 3 insertions(+)
diff --git a/src/glsl/Makefile.sources b/src/glsl/Makefile.sources
index d784a81..7bdd895 100644
Signed-off-by: Connor Abbott
---
src/glsl/nir/nir.c | 16
src/glsl/nir/nir.h | 2 ++
2 files changed, 18 insertions(+)
diff --git a/src/glsl/nir/nir.c b/src/glsl/nir/nir.c
index efc3f01..e3c37dd 100644
--- a/src/glsl/nir/nir.c
+++ b/src/glsl/nir/nir.c
@@ -2097,6 +2097,22 @@ nir
Signed-off-by: Connor Abbott
---
src/glsl/nir/nir_opt_dead_cf.c | 96 --
1 file changed, 84 insertions(+), 12 deletions(-)
diff --git a/src/glsl/nir/nir_opt_dead_cf.c b/src/glsl/nir/nir_opt_dead_cf.c
index 17eadbd..f96bf63 100644
--- a/src/glsl/nir/nir_opt
I'm not so sure about where to put the helper currently in nir.c... on
the one hand, it's pretty specific to this pass, but on the other hand
it needs to do some very fiddly low-level things to the control flow
which is why it needs access to a static function in nir.c
(stitch_blocks()) that I'd ra
Signed-off-by: Connor Abbott
---
src/glsl/nir/nir_opt_dead_cf.c | 126 ++---
1 file changed, 118 insertions(+), 8 deletions(-)
diff --git a/src/glsl/nir/nir_opt_dead_cf.c b/src/glsl/nir/nir_opt_dead_cf.c
index 3fbb794..17eadbd 100644
--- a/src/glsl/nir/nir_opt
This has started to happen more now that the backend is producing
KILL_IF more often.
Signed-off-by: Ilia Mirkin
---
.../drivers/nouveau/codegen/nv50_ir_peephole.cpp | 29 ++
.../nouveau/codegen/nv50_ir_target_nv50.cpp| 2 ++
2 files changed, 31 insertions(+)
diff
This covers the pattern where a KILL_IF is used, which triggers a
comparison of -x to 0. This can usually be folded into the comparison whose
result is being compared to 0, however it may, itself, have already been
combined with another comparison. That shouldn't impact the logic of
this pass howev
Signed-off-by: Ilia Mirkin
---
src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gk110.cpp | 12
src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gm107.cpp | 1 +
src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0.cpp | 8 +++-
3 files changed, 16 insertions(+), 5 deletions(
Signed-off-by: Ilia Mirkin
---
Pretty sure there's nothing wrong with it, but it looks odd in the code.
src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gk110.cpp | 2 ++
src/gallium/drivers/nouveau/codegen/nv50_ir_emit_gm107.cpp | 7 +--
src/gallium/drivers/nouveau/codegen/nv50_ir_emit_nvc0
Reviewed-by: Jason Ekstrand wrote:
> We were rewriting the uses of the intrinsic instruction to point to
> something else after removing it, which only happened to work since we
> were lax about having dangling uses when removing instructions. Fix that
> up.
>
> Signed-off-by: Connor Abbott
> ---
On Fri, May 8, 2015 at 10:03 PM, Connor Abbott wrote:
> The point of cleanup_defs_uses() is to make an instruction safe to
> remove by removing any references that the rest of the shader may have
> to it. Previously, it was handling register use/def sets and removing
> the instruction from the use
On Sat, May 9, 2015 at 1:52 AM, Jason Ekstrand wrote:
> On Fri, May 8, 2015 at 10:03 PM, Connor Abbott wrote:
>> The point of cleanup_defs_uses() is to make an instruction safe to
>> remove by removing any references that the rest of the shader may have
>> to it. Previously, it was handling regis
On Fri, May 8, 2015 at 10:03 PM, Connor Abbott wrote:
> This series implements a bunch of related optimizations that run at
> once as part of the same pass to eliminate control flow that is
> useless. This includes both unreachable code and useless loops, i.e.
> loops that don't compute a result u
88 matches
Mail list logo
