Signed-off-by: Samuel Pitoiset
---
src/gallium/drivers/nouveau/nvc0/nvc0_tex.c | 16 +++-
1 file changed, 3 insertions(+), 13 deletions(-)
diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c
b/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c
index
With this patch, if a .shader_test file contains
[require]
...
SSO ENABLED
then we'll set GL_PROGRAM_SEPARABLE to compile the shaders into separate
shader objects. This prevents the linker from removing unused inputs
and outputs. Drivers may also choose to lay out interfaces of SSO
There are a couple cases where a single shader might happen:
- compute shaders
(only one stage, no inputs and outputs; separable shouldn't matter)
- vertex shaders with transform feedback
(we want to retain outputs, but transform feedback varyings are
specified via the API, not the shader
Sounds reasonable. Pretty sure those NV_VRAM_DOMAIN thingies should
just be txc->domain. With that fixed,
Reviewed-by: Ilia Mirkin
On Mon, Jul 25, 2016 at 6:17 PM, Samuel Pitoiset
wrote:
> Signed-off-by: Samuel Pitoiset
On Mon, Jul 25, 2016 at 6:21 PM, Samuel Pitoiset
wrote:
>
>
> On 07/26/2016 12:20 AM, Ilia Mirkin wrote:
>>
>> Sounds reasonable. Pretty sure those NV_VRAM_DOMAIN thingies should
>> just be txc->domain. With that fixed,
>
>
> IIRC, this NV_VRAM_DOMAIN thing is for
This patche switches non-TGSI compute shaders over to using the HSA
ABI described here:
https://github.com/RadeonOpenCompute/ROCm-Docs/blob/master/AMDGPU-ABI.md
The HSA ABI provides a much cleaner interface for compute shaders and allows
us to share more code in the compiler with the HSA stack.
https://bugs.freedesktop.org/show_bug.cgi?id=96950
--- Comment #8 from Brian Paul ---
(In reply to Mathias Fröhlich from comment #6)
> Created attachment 125255 [details] [review]
> Fix for an other assert in immediate mode rendering with the linux turbine
>
> Brian,
>
> I
On Haswell:
total instructions in shared programs: 6193528 -> 6197476 (0.06%)
instructions in affected programs: 756099 -> 760047 (0.52%)
helped: 1057
HURT: 2983
total cycles in shared programs: 71120700 -> 71169938 (0.07%)
cycles in affected programs: 5378 -> 53493716 (0.09%)
helped: 12737
https://bugs.freedesktop.org/show_bug.cgi?id=97019
--- Comment #8 from Francisco Jerez ---
(In reply to Dieter Nützel from comment #6)
> (In reply to Francisco Jerez from comment #5)
> > Seems like a GCC bug... You may be able to work around the issue by using
> > the
On 07/26/2016 12:24 AM, Ilia Mirkin wrote:
On Mon, Jul 25, 2016 at 6:21 PM, Samuel Pitoiset
wrote:
On 07/26/2016 12:20 AM, Ilia Mirkin wrote:
Sounds reasonable. Pretty sure those NV_VRAM_DOMAIN thingies should
just be txc->domain. With that fixed,
IIRC, this
Signed-off-by: Samuel Pitoiset
---
src/gallium/drivers/nouveau/nvc0/nvc0_tex.c | 45 ++---
1 file changed, 9 insertions(+), 36 deletions(-)
diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c
b/src/gallium/drivers/nouveau/nvc0/nvc0_tex.c
On Fri, Jul 22, 2016 at 8:58 PM, Francisco Jerez wrote:
> This boolean flag was being used for two different things:
>
> - To set the brw_wm_prog_data::dual_src_blend flag. Instead we can
>just set it based on whether the dual_src_output register is valid,
>which
Reviewed-by: Ilia Mirkin
On Mon, Jul 25, 2016 at 6:48 PM, Samuel Pitoiset
wrote:
> Signed-off-by: Samuel Pitoiset
> ---
> src/gallium/drivers/nouveau/nvc0/nvc0_tex.c | 16 +++-
> 1 file changed, 3
On 07/26/2016 12:20 AM, Ilia Mirkin wrote:
Sounds reasonable. Pretty sure those NV_VRAM_DOMAIN thingies should
just be txc->domain. With that fixed,
IIRC, this NV_VRAM_DOMAIN thing is for gk20a, maybe this is going to fix
some tests as a side effect?
Reviewed-by: Ilia Mirkin
Rob Clark writes:
> On Mon, Jul 25, 2016 at 8:47 PM, Eric Anholt wrote:
>> Rob Herring writes:
>>
>>> It is necessary to reuse existing BOs when dmabufs are imported. There
>>> are 2 cases that need to be handled. dmabufs can be
Iago Toral Quiroga writes:
> From: Connor Abbott
>
> ---
> src/mesa/drivers/dri/i965/brw_vec4_nir.cpp | 5 -
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_vec4_nir.cpp
>
Rob Herring writes:
> It is necessary to reuse existing BOs when dmabufs are imported. There
> are 2 cases that need to be handled. dmabufs can be created/exported and
> imported by the same process and can be imported multiple times.
> Copying other drivers, add a hash table to
Iago Toral Quiroga writes:
> From: Connor Abbott
>
Reviewed-by: Francisco Jerez
> ---
> src/mesa/drivers/dri/i965/brw_vec4.cpp | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git
On Mon, Jul 25, 2016 at 1:19 PM, Marek Olšák wrote:
> On Mon, Jul 25, 2016 at 5:42 PM, Rob Clark wrote:
>> On Mon, Jul 25, 2016 at 11:16 AM, Brian Paul wrote:
>>> On 07/18/2016 07:11 AM, Marek Olšák wrote:
@@ -183,49 +107,42 @@
On Mon, 2016-07-25 at 16:54 -0700, Kenneth Graunke wrote:
> With this patch, if a .shader_test file contains
>
> [require]
> ...
> SSO ENABLED
>
> then we'll set GL_PROGRAM_SEPARABLE to compile the shaders into
> separate
> shader objects. This prevents the linker from removing
ping
On Sat, Jul 16, 2016 at 12:21 PM, Ilia Mirkin wrote:
> The additional provision of GL_OES_copy_image is that it work for ETC.
> However many desktop GPUs don't have native ETC support, so st/mesa does
> the decoding by hand. Instead of discarding the compressed data,
---
src/gallium/drivers/radeonsi/si_compute.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/gallium/drivers/radeonsi/si_compute.c
b/src/gallium/drivers/radeonsi/si_compute.c
index 5a40286..949ab1a 100644
--- a/src/gallium/drivers/radeonsi/si_compute.c
+++
On Mon, Jul 25, 2016 at 8:47 PM, Eric Anholt wrote:
> Rob Herring writes:
>
>> It is necessary to reuse existing BOs when dmabufs are imported. There
>> are 2 cases that need to be handled. dmabufs can be created/exported and
>> imported by the same process and
Iago Toral Quiroga writes:
Reviewed-by: Francisco Jerez
> ---
> src/mesa/drivers/dri/i965/brw_vec4_nir.cpp | 8
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/src/mesa/drivers/dri/i965/brw_vec4_nir.cpp
>
Iago Toral Quiroga writes:
> Basically, this involves considering the bit-size information to set
> the appropriate type on both operands and destination.
> ---
> src/mesa/drivers/dri/i965/brw_vec4_nir.cpp | 12
> 1 file changed, 8 insertions(+), 4 deletions(-)
>
Signed-off-by: Dieter Nützel
---
src/gallium/state_trackers/clover/llvm/codegen/native.cpp | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/state_trackers/clover/llvm/codegen/native.cpp
Iago Toral Quiroga writes:
> From: Connor Abbott
>
> Less duplication, one one less case to handle for doubles and support
> for sized NIR types.
>
> v2: Fix call to get_instance by swapping rows and columns params (Iago)
>
> Signed-off-by: Iago
On Mon, 2016-07-25 at 16:54 -0700, Kenneth Graunke wrote:
> There are a couple cases where a single shader might happen:
>
> - compute shaders
> (only one stage, no inputs and outputs; separable shouldn't matter)
> - vertex shaders with transform feedback
> (we want to retain outputs, but
Pushed, thanks!
On 22.07.2016 13:10, Nils Wallménius wrote:
When possible, do the memcpy on larger blocks. This reduces cycles
spent in _mesa_propagate_uniforms_to_driver_storage from
1.51 % to 0.62% according to perf during the Unigine Heaven benchmark.
It did not affect the framerate of the
Am 23.07.2016 um 01:51 schrieb Andy Furniss:
Christian König wrote:
From: Boyuan Zhang
For putimage call, if image format is yv12 (or IYUV with U V field swap)
This comment confuses me
AIUI + checking on fourcc.org yv12 is YVU and IYUV/I420 is YUV and
nv12 is
Apart for patch 5, the series is
Reviewed-by: Nicolai Hähnle
On 23.07.2016 02:14, Marek Olšák wrote:
Hi,
This is for GPU hangs that are hard to reproduce and require interactive
playing for minutes or even hours.
The performance should be at least 50% of the
Nayan Deshmukh wrote:
Thanks for testing :)
On Monday, July 25, 2016, Andy Furniss wrote:
Nayan Deshmukh wrote:
Hi Christian,
I have sent the new patches, they should fix all the artifacts. :)
I have briefly tried these over time and v3 1/2 + v2 2/2 still show
On 23.07.2016 02:14, Marek Olšák wrote:
From: Marek Olšák
---
src/gallium/drivers/ddebug/dd_screen.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/gallium/drivers/ddebug/dd_screen.c
b/src/gallium/drivers/ddebug/dd_screen.c
index
https://bugs.freedesktop.org/show_bug.cgi?id=93551
--- Comment #33 from Mikhail Korolev ---
Created attachment 125311
--> https://bugs.freedesktop.org/attachment.cgi?id=125311=edit
divos-hack.patch
(In reply to Thomas J. Moore from comment #32)
> Created attachment 125302
On 07/18/2016 07:11 AM, Marek Olšák wrote:
From: Marek Olšák
The goal is to do this in st_validate_state:
while (dirty)
atoms[u_bit_scan()]->update(st);
That implies that atoms can't specify which flags they consume.
There is exactly one ST_NEW_* flag for each
On Mon, Jul 25, 2016 at 11:16 AM, Brian Paul wrote:
> On 07/18/2016 07:11 AM, Marek Olšák wrote:
>>
>> From: Marek Olšák
>>
>> The goal is to do this in st_validate_state:
>> while (dirty)
>>atoms[u_bit_scan()]->update(st);
>>
>> That implies
On 07/22/2016 11:40 AM, Jose Fonseca wrote:
(2nd try. 1st email is being held due to size.)
On 21/07/16 18:48, Vedran Miletić wrote:
LLVM and Mesa both define the DEBUG macro in incompatible ways. As a
general practice, we should avoid using such generic names when it is
possible to do so.
Am 22.07.2016 um 17:21 schrieb Emil Velikov:
On 22 July 2016 at 09:42, Christian König wrote:
Am 22.07.2016 um 03:37 schrieb Rob Clark:
On Thu, Jul 21, 2016 at 9:35 PM, Rob Clark wrote:
On Thu, Jul 21, 2016 at 1:48 PM, Vedran Miletić
https://bugs.freedesktop.org/show_bug.cgi?id=93551
--- Comment #34 from Mikhail Korolev ---
(In reply to Thomas J. Moore from comment #32)
> Created attachment 125302 [details]
> Simple LD_PRELOAD shim to apply necessary patches for divos
>
> Game works great for me with the
On Fri, Jul 22, 2016 at 10:39 PM, Jason Ekstrand wrote:
> intel_mipmap_tree::logical_depth0 is now in 2-D slices so there is no need
> for us to multiply by 6 when we go to fill out a blorp surface state.
>
> Signed-off-by: Jason Ekstrand
> ---
>
On Thu, Jul 21, 2016 at 9:21 PM, Jason Ekstrand wrote:
> Signed-off-by: Jason Ekstrand
> Cc: "12.0"
> Cc: Nanley Chery
> ---
> src/intel/vulkan/anv_meta_clear.c | 6 --
> 1 file changed,
On 07/22/2016 10:22 AM, Rob Herring wrote:
Use the common pipe_screen ref counting and fd hashing functions. The
mutex can be dropped as the pipe loader protects the create_screen()
calls.
Signed-off-by: Rob Herring
---
src/gallium/auxiliary/target-helpers/drm_helper.h | 2
https://bugs.freedesktop.org/show_bug.cgi?id=90264
--- Comment #67 from Diego Viola ---
(In reply to Diego Viola from comment #66)
> I have this issue as well, I'm on Arch Linux (x86-64).
>
> mesa 12.0.1-2
> xorg-server 1.18.4-1
> linux 4.6.4-1
>
> Using modesetting
https://bugs.freedesktop.org/show_bug.cgi?id=93551
--- Comment #36 from Mikhail Korolev ---
(In reply to Thomas J. Moore from comment #35)
> (In reply to Mikhail Korolev from comment #33)
> > Your code calls dlsym at every call of glGetString/glXGetProcAddressARB
> > instead
From: Nicolai Hähnle
When an application declares varying arrays but does not actually do any
indirect indexing, some array indices may end up unused in the consuming
shader, so the number of input slots that correspond to the array ends
up less than the array_size.
Cc:
On Mon, Jul 25, 2016 at 4:30 PM, Andy Furniss wrote:
> Nayan Deshmukh wrote:
>
>> Thanks for testing :)
>>
>> On Monday, July 25, 2016, Andy Furniss wrote:
>>
>> Nayan Deshmukh wrote:
>>>
>>> Hi Christian,
I have sent the new patches, they
https://bugs.freedesktop.org/show_bug.cgi?id=93551
--- Comment #35 from Thomas J. Moore ---
(In reply to Mikhail Korolev from comment #33)
> Your code calls dlsym at every call of glGetString/glXGetProcAddressARB
> instead of call it only at startup.
If so, your version of
Rob Herring writes:
> Use the common pipe_screen ref counting and fd hashing functions for
> vc4. This is necessary to only create a single pipe_screen for a
> process and avoid multiple imports of same prime fd among other things
> (probably).
Reviewed-by: Eric Anholt
On Mon, Jul 25, 2016 at 5:42 PM, Rob Clark wrote:
> On Mon, Jul 25, 2016 at 11:16 AM, Brian Paul wrote:
>> On 07/18/2016 07:11 AM, Marek Olšák wrote:
>>>
>>> From: Marek Olšák
>>>
>>> The goal is to do this in st_validate_state:
>>>
https://bugs.freedesktop.org/show_bug.cgi?id=96950
--- Comment #7 from Rob Clark ---
(In reply to Mathias Fröhlich from comment #6)
> Created attachment 125255 [details] [review]
> Fix for an other assert in immediate mode rendering with the linux turbine
>
> Rob,
>
>
https://bugs.freedesktop.org/show_bug.cgi?id=90264
--- Comment #66 from Diego Viola ---
I have this issue as well, I'm on Arch Linux (x86-64).
mesa 12.0.1-2
xorg-server 1.18.4-1
linux 4.6.4-1
Using modesetting here, ThinkPad T450 (broadwell i5-5300U).
--
You are
On Fri, 2016-07-22 at 23:09 +0200, Pali Rohár wrote:
> Hello,
>
> after fixing problem with mako version mesa from git still fails to
> compile. Now problematic part is clover state tracker. Error message
> is:
you are using old compiler. check:
https://bugs.freedesktop.org/show_bug.cgi?id=97019
52 matches
Mail list logo