On 09/09/2015 04:04 AM, Kenneth Graunke wrote:
Our old value of 16384 is the minimum value. DirectX apparently
requires 65536 at a minimum; that's also what nVidia and the Intel
Windows driver advertise. AMD advertises MAX_INT.
Ilia Mirkin noticed that "Shadow Warrior" uses UBOs larger than
On 07.09.2015 07:14, Marek Olšák wrote:
> From: Marek Olšák
This patch is
Reviewed-by: Michel Dänzer
Patches 1 & 2 are
Acked-by: Michel Dänzer
--
Earthling Michel Dänzer | http://www.amd.com
On Thu, Sep 03, 2015 at 11:08:11AM -0700, Ian Romanick wrote:
> On 09/03/2015 09:05 AM, Chris Wilson wrote:
> > If the user supplies a pixel format of GL_RGB + GL_UNSIGNED_SHORT_5_6_5
> > and specifies a generic unsized GL_RGB internal format, match that to a
> > texture format of
This pass will propagate the destination components of a vecN instructions,
as destination of the instructions that define its sources; if certain
conditions are met.
If all the components of the destination register in the vecN instruction
can be propagated, the instruction is removed.
Enable the lower_vec_and_coalesce pass on non-scalar shaders, after
the shader is out of SSA-form and before passing lower_vec_to_movs.
shader-db VS results against non-NIR before this patch:
total instructions in shared programs: 1853747 -> 1804428 (-2.66%)
instructions in affected programs:
On 07.09.2015 07:17, Marek Olšák wrote:
> From: Marek Olšák
>
> We will only do depth-only or stencil-only decompress blits, whichever is
> needed by textures, instead of always doing both.
> ---
> src/gallium/drivers/r600/evergreen_state.c| 4 ++--
>
On 07.09.2015 07:17, Marek Olšák wrote:
> From: Marek Olšák
>
> instead of always doing both.
> Usually, only depth is needed, so stencil decompression is useless.
[...]
> @@ -247,6 +256,7 @@ void si_flush_depth_textures(struct si_context *sctx,
>
On Thu, Sep 3, 2015 at 6:05 PM, Chris Wilson wrote:
> If the user supplies a pixel format of GL_RGB + GL_UNSIGNED_SHORT_5_6_5
> and specifies a generic unsized GL_RGB internal format, match that to a
> texture format of MESA_FORMAT_B5G6R5 if supported by the hardware.
>
CB updates to bound buffers need to go through the CB_DATA endpoints,
otherwise the shader may not notice that the updates happened.
Furthermore, these updates have to go in to the same address as the
bound buffer, otherwise, again, the shader may not notice updates.
So we keep track of all the
On 08/09/15 19:50, Jordan Justen wrote:
> On 2015-08-05 01:30:21, Iago Toral Quiroga wrote:
>> From: Samuel Iglesias Gonsalvez
>>
>> Otherwise, generate a link time error as per the
>> ARB_shader_storage_buffer_object spec.
>>
>> Signed-off-by: Samuel Iglesias Gonsalvez
On 08/09/15 18:52, Jordan Justen wrote:
> On 2015-08-05 01:30:20, Iago Toral Quiroga wrote:
>> From: Samuel Iglesias Gonsalvez
>>
>> Signed-off-by: Samuel Iglesias Gonsalvez
>> ---
>> src/glsl/lower_ubo_reference.cpp | 65
>>
On 31/08/15 08:38, Samuel Iglesias Gonsálvez wrote:
>
>
> On 29/08/15 02:27, Jordan Justen wrote:
>> On 2015-08-05 01:30:17, Iago Toral Quiroga wrote:
>>> From: Samuel Iglesias Gonsalvez
>>>
>>> This commit also adds functions to calculate std430 base alignment and sizes
On Wed, Sep 9, 2015 at 11:25 AM, Chris Wilson wrote:
> On Wed, Sep 09, 2015 at 11:11:59AM +0200, Erik Faye-Lund wrote:
>> On Thu, Sep 3, 2015 at 6:05 PM, Chris Wilson
>> wrote:
>> > If the user supplies a pixel format of GL_RGB +
These functions will be used by textureQueryLod.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 100 +--
src/gallium/drivers/softpipe/sp_tex_sample.h | 7 ++
2 files changed, 101 insertions(+), 6 deletions(-)
diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c
It clearly is here by accident.
---
src/gallium/auxiliary/tgsi/tgsi_exec.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c
b/src/gallium/auxiliary/tgsi/tgsi_exec.c
index 75cd0d5..9544623 100644
---
textureQueryLod returns a vec2 with a mipmap information and a
LOD. The latter needs to be not clamped.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 55
1 file changed, 39 insertions(+), 16 deletions(-)
diff --git a/src/gallium/drivers/softpipe/sp_tex_sample.c
This introduces new vfunc in tgsi_sampler just for this opcode. I
decided against extending get_samples vfunc to return the mipmap level
and LOD - the function's prototype is already too scary and doing the
sampling for textureQueryLod would be a waste of time.
---
This function will be later used by textureQueryLod. The
img_filter_func are optional, because textureQueryLod will not need
them.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 53 +++-
1 file changed, 37 insertions(+), 16 deletions(-)
diff --git
The level-of-detail bias wasn't simply added in the explicit LOD case.
This case seems to be tested only in piglit's
fs-texturequerylod-nearest-biased test, which is currently skipped, as
softpipe does not support textureQueryLod at the moment.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 2
Passes the shader piglit tests and introduces no regressions.
This commit finally makes use of the refactoring in previous
commits.
---
src/gallium/drivers/softpipe/sp_screen.c | 2 +-
src/gallium/drivers/softpipe/sp_tex_sample.c | 47 +++-
2 files changed, 47
This is to avoid tying the conversion to sampling - textureQueryLod
will need to do the conversion too, but it does not do any sampling.
So instead of a "sample" vfunc, there is a "convert" vfunc. The
drawback of this approach is that a "noop" convert copies 3 arrays of
floats. Would be nice to
Putting this function pointer into a struct enables grouping of
several related functions in a single place. For now it is just a
single function, but the struct will be later extended with a
mip_level_func for returning mip level.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 28
The following patches do some preparatory refactoring in softpipe's
sp_tex_sample.{c,h} before actually implementing the feature.
There is also a patch that fixes softpipe's textureLod with bias. I
caught this bug when finished the textureQueryLod implementation - one
of four textureQueryLod
Comparison with a signed expression and unsigned value
is converted to unsigned value, reason for minus value is interpreted
as a big unsigned value. For this case the "for" loop
is going into unexpected behavior.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38109
Signed-off-by: Marius
Mark Janes writes:
> Francisco Jerez writes:
>
>> Mark Janes writes:
>>
>>> When I tested this, I saw an intermittent BSW gpu hang. I haven't been
>>> able to confirm that it is due to the host-mem-barrier test.
>>>
>> It
On Wed, Sep 09, 2015 at 11:11:59AM +0200, Erik Faye-Lund wrote:
> On Thu, Sep 3, 2015 at 6:05 PM, Chris Wilson wrote:
> > If the user supplies a pixel format of GL_RGB + GL_UNSIGNED_SHORT_5_6_5
> > and specifies a generic unsized GL_RGB internal format, match that to a
>
https://bugs.freedesktop.org/show_bug.cgi?id=80821
Emil Velikov changed:
What|Removed |Added
Status|NEW |RESOLVED
On Wed, Sep 09, 2015 at 12:09:40PM +0200, Erik Faye-Lund wrote:
> On Wed, Sep 9, 2015 at 11:25 AM, Chris Wilson
> wrote:
> > On Wed, Sep 09, 2015 at 11:11:59AM +0200, Erik Faye-Lund wrote:
> >> On Thu, Sep 3, 2015 at 6:05 PM, Chris Wilson
>
Ian Romanick writes:
> On 09/03/2015 06:03 AM, Francisco Jerez wrote:
>> IVB and VLV hang sporadically when an untyped surface read or write
>> message is used to access a surface of format other than RAW, as may
>> happen when there is a mismatch between the format
On Wed, 9 Sep 2015 09:11:50 -0400
Alex Deucher wrote:
> > Oh, absolutely - I had no issues with "this needs changing". My issue
> > was with the fact it took months to get that. Had I come up with a new
> > patch, it would likely have taken a similar time, months again,
Am 09.09.2015 um 15:04 schrieb Roland Scheidegger:
> Am 09.09.2015 um 05:56 schrieb Ian Romanick:
>> On 08/15/2015 02:24 AM, Francisco Jerez wrote:
>>> Roland Scheidegger writes:
>>>
I guess though you'd need these bits when implementing things like
Some modern apps try to use msaa without keeping in mind the
restrictions on videomem of older cards. Resulting in dmesg saying:
[ 1197.850642] nouveau E[soffice.bin[3785]] fail ttm_validate
[ 1197.850648] nouveau E[soffice.bin[3785]] validating bo list
[ 1197.850654] nouveau
We do not have a generic blitter on nv3x cards, so we must use the
sifm object for color resolving.
This commit divides the sources and dest surfaces in to tiles which
match the constraints of the sifm object, so that color resolving
will work properly on nv3x cards.
Signed-off-by: Hans de Goede
Since _mesa_update_state() may be called several times per draw call,
we wish to be judicious about the amount of work we place inside the
intel_update_state() callback. By moving the current HiZ/color resolves
we need before drawing from out of the notify and into the draw itself,
we can save a
Hi,
On 08-09-15 09:53, Hans de Goede wrote:
Hi,
On 07-09-15 21:55, Ilia Mirkin wrote:
May I ask why you're doing 512x512 instead of 1024x1024? These are
already scaled up coordinates, so 1024x1024 should work no? Or is it
because of the seams on the edges? Do those not also appear with
On Wed, Sep 09, 2015 at 04:07:09PM +0200, Michał Winiarski wrote:
> From: Chris Wilson
>
> Userspace can pass in an offset that it presumes the object is located
> at. The kernel will then do its utmost to fit the object into that
> location. The assumption is that
Comparison with a signed expression and unsigned value
is converted to unsigned value, reason for minus value is interpreted
as a big unsigned value. For this case the "for" loop
is going into unexpected behavior.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=38109
Signed-off-by: Marius
Having checked whether the base class (gl_texture_object) is NULL, we
know that intel_texture_object itself cannot be NULL.
Signed-off-by: Chris Wilson
Cc: Jordan Justen
Cc: Jason Ekstrand
Cc: Kenneth Graunke
On 09/09/2015 07:59 AM, Marius Predut wrote:
On PNV platform, for 1 pixel line thickness or less,
the general anti-aliasing algorithm gives up, and a garbage line is generated.
Setting a Line Width of 0.0 specifies the rasterization
of the "thinnest" (one-pixel-wide), non-antialiased lines.
Performing the HiZ/color resolves is a fairly heavyweight operation that
we don't need to perform on every state update, only before use the
unresolved surfaces in a draw call. Since we use meta operations to
perform the resolve, we cannot do so from inside brw_draw_prims() itself
(as that will
When we come to flush the vertices, we need to clear the NeedFlush
flag first before calling into the backend as that backend may trigger
FLUSH_VERTICES() itself (through use of meta) before dirtying the vertex
state. If the NeedFlush flag is still in place, we end up recursing
into the backend
Now that we moved the resolving from out of the _mesa_update_state()
callback we can remove the trick of calling _mesa_update_state() again
from inside the resolver (as we no longer need to adhere to the
_mesa_update_state() contract). However, we still need to make sure that
the context is up to
A common problem with using HiZ and multisampling is that surfaces need
to resolved prior to use. Currently i965 does this inside its state
update hook, but that is a comparatively heavyweight operation that need
not be performed so frequently. The obvious solution (and therefore
fraught with
The very first operation performed by _mesa_readpixels() is to call
_mesa_update_state() - we do not need to do so ourselves.
Signed-off-by: Chris Wilson
Cc: Jordan Justen
Cc: Jason Ekstrand
Cc: Kenneth Graunke
Only walk through the set of enabled TextureUnits looking for a texture
that needs to be resolved if the context state flags a new texture.
Note that this will miss if the client is rendering into a texture that
it is reading from, though that needs explicit barriers (and futhermore
no piglits
On PNV platform, for 1 pixel line thickness or less,
the general anti-aliasing algorithm gives up, and a garbage line is generated.
Setting a Line Width of 0.0 specifies the rasterization
of the "thinnest" (one-pixel-wide), non-antialiased lines.
Lines rendered with zero Line Width are rasterized
On Sun, Sep 6, 2015 at 4:28 AM, Lauri Kasanen wrote:
> On Sat, 5 Sep 2015 23:29:05 +
> Albert Freeman wrote:
>
>> The reply from Eric Anholt made two suggestions that should not be
>> difficult to implement for someone who made the patch in the first
Am 09.09.2015 um 05:56 schrieb Ian Romanick:
> On 08/15/2015 02:24 AM, Francisco Jerez wrote:
>> Roland Scheidegger writes:
>>
>>> I guess though you'd need these bits when implementing things like
>>> ARB_fragment_shader_ordering? (Thus stuff actually looks useful but I
>>>
On 09/09/2015 08:16 AM, Marius Predut wrote:
Comparison with a signed expression and unsigned value
is converted to unsigned value, reason for minus value is interpreted
as a big unsigned value. For this case the "for" loop
is going into unexpected behavior.
Bugzilla:
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
Passes the shader piglit tests and introduces no regressions.
This commit finally makes use of the refactoring in previous
commits.
---
src/gallium/drivers/softpipe/sp_screen.c | 2 +-
src/gallium/drivers/softpipe/sp_tex_sample.c | 47
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
This introduces new vfunc in tgsi_sampler just for this opcode. I
decided against extending get_samples vfunc to return the mipmap level
and LOD - the function's prototype is already too scary and doing the
sampling for textureQueryLod would be a
On Wed, Sep 9, 2015 at 9:52 AM, Hans de Goede wrote:
> We do not have a generic blitter on nv3x cards, so we must use the
> sifm object for color resolving.
>
> This commit divides the sources and dest surfaces in to tiles which
> match the constraints of the sifm object, so
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
The following patches do some preparatory refactoring in softpipe's
sp_tex_sample.{c,h} before actually implementing the feature.
There is also a patch that fixes softpipe's textureLod with bias. I
caught this bug when finished the textureQueryLod
On Wed, Sep 9, 2015 at 9:52 AM, Hans de Goede wrote:
> Some modern apps try to use msaa without keeping in mind the
> restrictions on videomem of older cards. Resulting in dmesg saying:
>
> [ 1197.850642] nouveau E[soffice.bin[3785]] fail ttm_validate
> [ 1197.850648]
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
Putting this function pointer into a struct enables grouping of
several related functions in a single place. For now it is just a
single function, but the struct will be later extended with a
mip_level_func for returning mip level.
---
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
textureQueryLod returns a vec2 with a mipmap information and a
LOD. The latter needs to be not clamped.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 55
1 file changed, 39 insertions(+), 16 deletions(-)
diff
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
This function will be later used by textureQueryLod. The
img_filter_func are optional, because textureQueryLod will not need
them.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 53 +++-
1 file changed, 37
Softpin allows userspace to take greater control of GPU virtual address
space and eliminates the need of relocations. It can also be used to
mirror addresses between GPU and CPU (shared virtual memory).
Calls to drm_intel_bo_emit_reloc are still required to build the list of
While support for 48b ppgtt is here, parameter enabling it is not known
to the sanitize function. Let's update it to allow selecting
full_48bit_ppgtt using module parameter.
Cc: Michel Thierry
Cc: Mika Kuoppala
Signed-off-by: Michał Winiarski
The goal of this series is to start a discussion whether the interface and
implementation of softpin support proposed for libdrm is acceptable by all
interested parties.
The i915 patches are added so that it's easy to apply the series on latest
drm-intel-nightly and libdrm master and start using
From: Chris Wilson
Userspace can pass in an offset that it presumes the object is located
at. The kernel will then do its utmost to fit the object into that
location. The assumption is that userspace is handling its own object
locations (for example along with
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
These functions will be used by textureQueryLod.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 100 +--
src/gallium/drivers/softpipe/sp_tex_sample.h | 7 ++
2 files changed, 101 insertions(+), 6 deletions(-)
diff
On 09/09/2015 04:35 AM, Krzesimir Nowak wrote:
This is to avoid tying the conversion to sampling - textureQueryLod
will need to do the conversion too, but it does not do any sampling.
So instead of a "sample" vfunc, there is a "convert" vfunc. The
drawback of this approach is that a "noop"
Jordan Justen writes:
> first_non_payload_grf may be updated in assign_urb_setup for FS or
> assign_vs_urb_setup for VS.
Patch looks good, but I think we should clarify in the commit message
that we're doing this so that compute will get first_non_payload_grf set
up
Instead of storing insertion points, we simply store a single stack of
control flow nodes representing the nir_if / nir_loop we're currently
filling in.
When handling ELSE, we can simply switch to the current nir_if's
else_list. When handling ENDIF or ENDLOOP, we move the cursor to
after the
From: Rob Clark
Sometimes a useful thing for compilers (or, for example, tgsi_to_nir) to
know. And pretty trivial for scan to figure this out for us.
Signed-off-by: Rob Clark
---
src/gallium/auxiliary/tgsi/tgsi_scan.c | 16
On Wednesday, September 09, 2015 06:31:06 PM Rob Clark wrote:
> From: Rob Clark
>
> Sometimes a useful thing for compilers (or, for example, tgsi_to_nir) to
> know. And pretty trivial for scan to figure this out for us.
>
> Signed-off-by: Rob Clark
Jordan Justen writes:
> Signed-off-by: Jordan Justen
Looks good - maybe change the subject to
i965: Set up sampler state for compute shaders
since this really only sets up that part of the state.
Reviewed-by: Kristian Høgsberg
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
On Wed, Sep 09, 2015 at 11:53:54AM -0700, Ian Romanick wrote:
> On 09/09/2015 11:16 AM, Marius Predut wrote:
> > Comparison with a signed expression and unsigned value
> > is converted to unsigned value, reason for minus value is interpreted
> > as a big unsigned value. For this case the "for"
Jordan Justen writes:
> Enable barrier in MEDIA_INTERFACE_DESCRIPTOR if the program uses the
> barrier() GLSL function.
>
> On Ivy Bridge and Haswell, this allows the piglit test
> tests/spec/arb_compute_shader/execution/simple-barrier-atomics.shader_test
> to pass. On
Mark Janes writes:
> Francisco Jerez writes:
>
>> Mark Janes writes:
>>
>>> When I tested this, I saw an intermittent BSW gpu hang. I haven't been
>>> able to confirm that it is due to the host-mem-barrier test.
>>>
>> It
Jordan Justen writes:
> Signed-off-by: Jordan Justen
>
Reviewed-by: Kristian Høgsberg
> ---
> src/mesa/drivers/dri/i965/brw_context.h | 2 +-
> src/mesa/drivers/dri/i965/brw_state_upload.c | 2 ++
> 2 files
On Wed, Sep 9, 2015 at 11:13 AM, Ian Romanick wrote:
> From: Ian Romanick
>
> Previously the result of the complicated clamp() expression just dropped
> on the floor: clamp does not modify any of its parameters. Looking at
> the surrounding code,
While we're here, we also fix up a couple of potential ralloc-parenting
bugs. In particular, we were calling nir_src_copy with the shader as the
mem context instead of the instruction.
---
src/glsl/nir/nir_lower_vec_to_movs.c | 22 --
1 file changed, 16 insertions(+), 6
Previously, we did this thing with keeping track of a separate start_idx
which was different from the iteration variable. I think this was a relic
of the way that GLSL IR implements writemasks. In NIR, if a given bit in
the writemask is unset then that channel is just "unused", not missing. In
We made this switch in the FS backend some time ago and it seems to make a
number of things a bit easier.
---
src/mesa/drivers/dri/i965/brw_nir.c| 2 +-
src/mesa/drivers/dri/i965/brw_vec4.h | 1 +
src/mesa/drivers/dri/i965/brw_vec4_nir.cpp | 21 ++---
3 files
---
src/glsl/nir/nir_from_ssa.c | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git a/src/glsl/nir/nir_from_ssa.c b/src/glsl/nir/nir_from_ssa.c
index 084f43d..eaf883d 100644
--- a/src/glsl/nir/nir_from_ssa.c
+++ b/src/glsl/nir/nir_from_ssa.c
@@ -512,9 +512,7 @@
On Wed, Sep 9, 2015 at 6:19 PM, Ian Romanick wrote:
> On 09/09/2015 04:55 PM, Anuj Phogat wrote:
>> On Wed, Sep 9, 2015 at 11:30 AM, Ian Romanick
wrote:
>>> I'm pretty sure our implementation of this extension is complete
>>> rubbish. I have attached
On 09/09/2015 03:32 AM, Samuel Iglesias Gonsálvez wrote:
> On 31/08/15 08:38, Samuel Iglesias Gonsálvez wrote:
>> On 29/08/15 02:27, Jordan Justen wrote:
>>> Hmm, you quoted the extension spec, but the 4.30 spec has: "When using
>>> the std430 storage layout, shader storage blocks will be laid out
On Wed, Sep 9, 2015 at 11:30 AM, Ian Romanick wrote:
> I'm pretty sure our implementation of this extension is complete
> rubbish. I have attached an image from the blit-scaled test. This
> result cannot be useful to anyone.
>
Resending it with link to the image in place
---
src/glsl/nir/nir.c | 24
src/glsl/nir/nir.h | 2 ++
2 files changed, 26 insertions(+)
diff --git a/src/glsl/nir/nir.c b/src/glsl/nir/nir.c
index e173b21..0090b08 100644
--- a/src/glsl/nir/nir.c
+++ b/src/glsl/nir/nir.c
@@ -1175,6 +1175,30 @@
As of a10d4937, we would really like things associated with an instruction
to be allocated out of that instruction and not out of the shader. In
particular, you should be passing the instruction that will ultimately be
holding the source into nir_src_copy rather than an arbitrary memory
context.
---
src/glsl/nir/nir.c| 2 +-
src/glsl/nir/nir.h| 2 +-
src/glsl/nir/nir_control_flow.c | 2 +-
src/glsl/nir/nir_from_ssa.c | 5 ++---
src/glsl/nir/nir_lower_alu_to_scalar.c| 6 ++
---
src/glsl/nir/nir_lower_vec_to_movs.c | 18 --
1 file changed, 16 insertions(+), 2 deletions(-)
diff --git a/src/glsl/nir/nir_lower_vec_to_movs.c
b/src/glsl/nir/nir_lower_vec_to_movs.c
index e297cb8..7d31e36 100644
--- a/src/glsl/nir/nir_lower_vec_to_movs.c
+++
This is labeled HACK because it relies on the validator not validating
destination write-masks with respect to nir_op_infos[op].output_size and
relies on the fact that the i965 vec4 hardware does an implicit splat of
any dot-product results. Technically, nir_op_fdotN produces a single
component
The old pass blindly inserted a bunch of moves into the shader with no
concern for whether or not it was really needed. This adds code to try and
coalesce into the destination of the instruction providing the value.
Shader-db results for vec4 shaders on Haswell:
total instructions in shared
---
src/glsl/nir/nir.c | 15 ---
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/src/glsl/nir/nir.c b/src/glsl/nir/nir.c
index 9a22fea..e173b21 100644
--- a/src/glsl/nir/nir.c
+++ b/src/glsl/nir/nir.c
@@ -714,9 +714,16 @@ nir_instr_insert(nir_cursor cursor, nir_instr
This patch series started out with me suggesting to Eduardo that his
coalescing pass would be easier if the shader was in partial SSA. So I
started working towards getting the vec4 backend using partial SSA like the
FS backend does. Then I started adding more stuff to lower_vec_to_movs and
the
On 09/09/2015 04:55 PM, Anuj Phogat wrote:
> On Wed, Sep 9, 2015 at 11:30 AM, Ian Romanick wrote:
>> I'm pretty sure our implementation of this extension is complete
>> rubbish. I have attached an image from the blit-scaled test. This
>> result cannot be useful to anyone.
On Thursday, August 27, 2015 11:48:29 PM Ilia Mirkin wrote:
> This should include everything. I sent a test for textureSamples to
> piglit a while ago, not sure how to test imageSamples -- apparently ms
> images aren't entirely supported on i965? But I'm not sure via what
> feat that happens.
>
>
On 08/23/2015 01:50 AM, Jordan Justen wrote:
> We initialize gl_LocalInvocationIndex based on the extension spec
> formula:
>
> gl_LocalInvocationIndex =
> gl_LocalInvocationID.z * gl_WorkGroupSize.x * gl_WorkGroupSize.y +
> gl_LocalInvocationID.y * gl_WorkGroupSize.x +
>
On 09/07/2015 06:54 AM, Jose Fonseca wrote:
> On 07/09/15 10:17, Jean-Sébastien Pédron wrote:
>> On 04.09.2015 01:37, Matt Turner wrote:
>>> You need to test for this support in configure.ac. It's as simple as
>>> adding a call to AX_GCC_FUNC_ATTRIBUTE in the existing alphabetized
>>> list and
From: Ian Romanick
The target parameter of compute_msaa_layout appears to be unused since
83b83fb when support for CMS textures was added for Gen7.
The brw parameter of intel_get_non_msrt_mcs_alignment appears to be
unused since e92fbdc when the GEN check (along with
From: Ian Romanick
These only occurred in release builds, but they occurred in every file
that included intel_batchbuffer.h. Lots of spam. :(
intel_batchbuffer.h: In function 'intel_batchbuffer_advance':
intel_batchbuffer.h:153:47: warning: unused parameter 'brw'
From: Ian Romanick
intel_mipmap_tree.c: In function 'old_intel_miptree_unmap_raw':
intel_mipmap_tree.c:726:51: warning: unused parameter 'intel'
[-Wunused-parameter]
intel_miptree_unmap_raw(struct intel_context *intel,
From: Ian Romanick
intel_mipmap_tree.c: In function 'intel_miptree_unmap_gtt':
intel_mipmap_tree.c:777:34: warning: unused parameter 'map' [-Wunused-parameter]
struct intel_miptree_map *map,
^
intel_mipmap_tree.c:778:17: warning:
From: Ian Romanick
intel_fbo.c: In function 'intel_alloc_window_storage':
intel_fbo.c:415:48: warning: unused parameter 'ctx' [-Wunused-parameter]
intel_alloc_window_storage(struct gl_context * ctx, struct gl_renderbuffer *rb,
From: Ian Romanick
This hasn't been used outside intel_mipmap_tree.c since d5d4ba9 started
using meta instead of the blitter for PBO TexSubImage. While we're here,
1. s/intel_miptree_\([^_]\)_raw/miptree_\1_raw/g because we don't
generally prefix static functions.
2.
From: Ian Romanick
These only occurred in release builds, but they occurred in every file
that included intel_mipmap_tree.h. Lots of spam. :(
intel_mipmap_tree.h: In function 'intel_miptree_check_level_layer':
intel_mipmap_tree.h:595:59: warning: unused parameter 'mt'
1 - 100 of 147 matches
Mail list logo