We can't just break for padding slots. Instead, treat them like
unwritten output variables, so we handle flushing and incrementing
urb_offset correctly.
Paul introduced the concept of padding slots back in 2011, but we've
never actually used them for anything. So it's unsurprising that the
scala
Because we only support geometry shaders in core profile, we can safely
ignore any driver-extending of VS outputs.
Those are:
- Legacy userclipping (doesn't exist in core profile)
- Edgeflag copying (Gen4-5 only, no GS support)
- Point coord replacement (Gen4-5 only, no GS support)
- front/back co
Nothing actually relies on unused slots being initialized to
BRW_VARYING_SLOT_COUNT. Soon, we're going to have VUE maps with holes
in them, at which point pre-filling with BRW_VARYING_SLOT_PAD make a lot
more sense.
Signed-off-by: Kenneth Graunke
---
src/mesa/drivers/dri/i965/brw_vue_map.c | 2
Ilia Mirkin writes:
> On Fri, Sep 11, 2015 at 10:45 PM, Ian Romanick wrote:
>> On 09/11/2015 10:55 AM, Ilia Mirkin wrote:
>>> On Fri, Sep 11, 2015 at 12:36 PM, Marius Predut
>>> wrote:
Comparison with a signed expression and unsigned value
is converted to unsigned value, reason for m
On Fri, Sep 11, 2015 at 10:45 PM, Ian Romanick wrote:
> On 09/11/2015 10:55 AM, Ilia Mirkin wrote:
>> On Fri, Sep 11, 2015 at 12:36 PM, Marius Predut
>> wrote:
>>> Comparison with a signed expression and unsigned value
>>> is converted to unsigned value, reason for minus value is interpreted
>>>
On 09/11/2015 10:55 AM, Ilia Mirkin wrote:
> On Fri, Sep 11, 2015 at 12:36 PM, 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 g
For certain platforms that support rotated scanout buffers, currently,
there is no way to create them with the GBM DRI interface. This flag
will tell the DRI driver to set Y-tiling while creating the rotated
scanout buffer.
Cc: Matthew D Roper
Signed-off-by: Vivek Kasireddy
---
include/GL/inter
On 09/11/2015 03:41 AM, Predut, Marius wrote:
>
>
>> -Original Message-
>> From: Ian Romanick [mailto:i...@freedesktop.org]
>> Sent: Wednesday, September 09, 2015 8:54 PM
>> To: Predut, Marius; mesa-dev@lists.freedesktop.org
>> Subject: Re: [Mesa-dev] [PATCH v2] i915: fixing driver crashe
On Fri, Sep 11, 2015 at 7:54 PM, Ilia Mirkin wrote:
> On Mon, Sep 7, 2015 at 8:15 AM, Julien Isorce wrote:
>> - split nvc0_decoder_bsp in begin/next/end
>> - preserve content buffer when calling nvc0_decoder_bsp_next
>> - implement pipe_video_codec::begin_frame/end_frame
>>
>> https://bugs.freede
On Mon, Sep 7, 2015 at 8:15 AM, Julien Isorce wrote:
> - split nvc0_decoder_bsp in begin/next/end
> - preserve content buffer when calling nvc0_decoder_bsp_next
> - implement pipe_video_codec::begin_frame/end_frame
>
> https://bugs.freedesktop.org/show_bug.cgi?id=89969
>
> Signed-off-by: Julien Is
Given that coalescing seems to be hard, I'm bumping this series from
RFC to PATCH. It's kinda nuts, but seems fairly effective.
--Jason
On Tue, Sep 8, 2015 at 5:35 PM, Jason Ekstrand wrote:
> ---
> src/glsl/Makefile.sources| 1 +
> src/glsl/nir/nir.h
On 2015-09-10 05:48:50, Alejandro Piñeiro wrote:
> Not a full review, but a comment. See inline.
>
> On 23/08/15 09:09, Jordan Justen wrote:
> > We initialize gl_GlobalInvocationID based on the extension spec
> > formula:
> >
> > gl_GlobalInvocationID =
> > gl_WorkGroupID * gl_WorkGrou
On 09/11/2015 04:49 PM, Krzesimir Nowak wrote:
On Fri, Sep 11, 2015 at 11:57 PM, Emil Velikov mailto:emil.l.veli...@gmail.com>> wrote:
On 11 September 2015 at 19:07, Krzesimir Nowak mailto:krzesi...@kinvolk.io>> wrote:
> With that, sp_sampler_view instances are not abused anymore as a lo
On Fri, Sep 11, 2015 at 11:57 PM, Emil Velikov
wrote:
> On 11 September 2015 at 19:07, Krzesimir Nowak
> wrote:
> > With that, sp_sampler_view instances are not abused anymore as a local
> > storage, so we can later make them constant.
> > ---
> > src/gallium/drivers/softpipe/sp_tex_sample.c |
On Fri, Sep 11, 2015 at 11:33:14AM +0300, Abdiel Janulgue wrote:
> Some updated figures first:
>
> Ue4 Reflections Subway demo
> x fs gather constants disabled
> + fs gather constants enabled
>
> HSW:
> N Min MaxMedian AvgStddev
> x 10 5.
Signed-off-by: Ilia Mirkin
---
src/gallium/docs/source/screen.rst | 1 +
src/gallium/drivers/freedreno/freedreno_screen.c | 1 +
src/gallium/drivers/i915/i915_screen.c | 1 +
src/gallium/drivers/ilo/ilo_screen.c | 1 +
src/gallium/drivers/llvmpipe/lp_screen.c
The image component of the ext is a no-op since there is no image support
in gallium (yet).
Signed-off-by: Ilia Mirkin
Reviewed-by: Edward O'Callaghan
Reviewed-by: Brian Paul
---
docs/GL3.txt | 2 +-
docs/relnotes/11.1.0.html | 2 +-
src/mesa/stat
OK, I've added a PIPE_CAP to control it now. I've left the relevant
R-b's from v1 in place despite a few changes, I think they're minor
enough.
Ilia Mirkin (5):
tgsi: add a TXQS opcode to retrieve the number of texture samples
gallium: add PIPE_CAP_TGSI_TXQS to let st know if TXQS is supported
Signed-off-by: Ilia Mirkin
Reviewed-by: Glenn Kennard
---
src/gallium/drivers/r600/r600_pipe.c | 2 +-
src/gallium/drivers/r600/r600_shader.c | 16
2 files changed, 13 insertions(+), 5 deletions(-)
diff --git a/src/gallium/drivers/r600/r600_pipe.c
b/src/gallium/drivers/r600
Signed-off-by: Ilia Mirkin
Reviewed-by: Roland Scheidegger
Reviewed-by: Edward O'Callaghan
---
src/gallium/auxiliary/tgsi/tgsi_info.c | 3 ++-
src/gallium/docs/source/tgsi.rst | 12 +++-
src/gallium/include/pipe/p_shader_tokens.h | 1 +
3 files changed, 14 insertions(+),
Signed-off-by: Ilia Mirkin
---
src/gallium/drivers/nouveau/codegen/nv50_ir.h | 4 ++--
.../drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp | 22 +-
.../nouveau/codegen/nv50_ir_lowering_nv50.cpp | 20
src/gallium/drivers/nouveau/nv50/nv50_screen.c
On Thu, 2015-09-10 at 21:36 -0700, Jason Ekstrand wrote:
> On Thu, Sep 10, 2015 at 7:43 PM, Timothy Arceri <
> t_arc...@yahoo.com.au> wrote:
> > On Tue, 2015-09-08 at 14:14 -0700, Jason Ekstrand wrote:
> > > On Mon, Sep 7, 2015 at 3:09 PM, Timothy Arceri <
> > > t_arc...@yahoo.com.au
> > > > wrote:
On Fri, Aug 28, 2015 at 7:50 AM, Nanley Chery wrote:
> From: Nanley Chery
>
> According to the extensions table and our glext headers,
> OES_compressed_ETC1_RGB8_texture is only supported in
> GLES1 and GLES2.
>
> Signed-off-by: Nanley Chery
> ---
> src/mesa/main/teximage.c | 3 ++-
> 1 file c
On 11 September 2015 at 19:07, Krzesimir Nowak wrote:
> With that, sp_sampler_view instances are not abused anymore as a local
> storage, so we can later make them constant.
> ---
> src/gallium/drivers/softpipe/sp_tex_sample.c | 36
> +---
> src/gallium/drivers/softpipe/s
Honestly, I don't see much point of constifying simple one-time-assigned
local variables (won't make a difference to the compiler anyway, and
even for reading imho doesn't add much). I don't object changing it
though, and the rest of the series definitely looks good to me, I'm glad
you got rid of t
On 09/11/2015 12:07 PM, Krzesimir Nowak wrote:
This stops abusing sp_sampler_view as a local storage and thus
allowing to put some more const modifiers in front of some types.
I might have gotten carried away too bit with the last commit. :) So
if you don't like it, then please ignore it.
Teste
On Fri, Aug 28, 2015 at 9:09 AM, Nanley Chery wrote:
> From: Nanley Chery
>
> Determines if a compressed format is non-palettized and specific.
>
> Signed-off-by: Nanley Chery
> ---
> src/mesa/main/texcompress.c | 10 ++
> src/mesa/main/texcompress.h | 3 +++
> 2 files changed, 13 ins
On Fri, Sep 11, 2015 at 12:12:15PM -0700, Jordan Justen wrote:
> On 2015-09-10 16:59:12, Ben Widawsky wrote:
> > All SKL SKUs except the lowest one which has half the L3 size actually have
> > 384K
>
> These commit message lines seem to wrap a bit long. This first line is
> 80 characters.
>
> >
On 2015-09-10 16:59:12, Ben Widawsky wrote:
> All SKL SKUs except the lowest one which has half the L3 size actually have
> 384K
These commit message lines seem to wrap a bit long. This first line is
80 characters.
> of URB per slice.
>
> For once, I can explain how this mistake was made and ho
On Fri, Sep 11, 2015 at 11:37:21AM -0700, Ben Widawsky wrote:
> On Fri, Sep 11, 2015 at 10:24:29AM -0700, Ben Widawsky wrote:
> > On Sun, Sep 06, 2015 at 06:12:38PM +0200, Francisco Jerez wrote:
> > > This series implements dynamic partitioning of the L3 cache space
> > > among its clients, the pur
On Fri, Sep 11, 2015 at 10:24:29AM -0700, Ben Widawsky wrote:
> On Sun, Sep 06, 2015 at 06:12:38PM +0200, Francisco Jerez wrote:
> > This series implements dynamic partitioning of the L3 cache space
> > among its clients, the purpose is multiple:
> >
> > - Steal a chunk of L3 space when necessary
On Fri, Sep 11, 2015 at 10:42 AM, Brian Paul wrote:
> On 09/10/2015 09:15 PM, Ilia Mirkin wrote:
>>
>> Signed-off-by: Ilia Mirkin
>> ---
>> src/gallium/auxiliary/tgsi/tgsi_info.c | 3 ++-
>> src/gallium/docs/source/tgsi.rst | 12 +++-
>> src/gallium/include/pipe/p_shade
On Fri, Sep 11, 2015 at 10:47 AM, Brian Paul wrote:
> On 09/11/2015 07:26 AM, Roland Scheidegger wrote:
>>
>> Am 11.09.2015 um 05:15 schrieb Ilia Mirkin:
>>>
>>> My hope was (as you can see in the last patch) to enable this for all
>>> drivers that support MS textures. I've got nv50/nvc0/r600g cov
This commit makes a lot of variables constant - this is basically done
by moving the computation to variable definition. Some of them are
moved into lower scopes (like in img_filter_2d_ewa).
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 441 ---
1 file changed, 199 ins
Those functions actually could always take them as constants.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 32 ++--
src/gallium/drivers/softpipe/sp_tex_sample.h | 4 ++--
2 files changed, 18 insertions(+), 18 deletions(-)
diff --git a/src/gallium/drivers/softpipe/sp
Add a small inline function doing the casting - this is to make sure
we don't do a cast from some completely unrelated type. This commit
does not make tgsi_sampler parameters const in vfuncs themselves for
now - probably llvmpipe would need looking at before making such a
change.
---
src/gallium/d
A followup from previous commit - since all functions called by
query_lod take pointers to const sp_sampler_view and const sp_sampler,
which are taken from tgsi_sampler subclass, we can the tgsi_sampler as
const itself now.
---
src/gallium/auxiliary/tgsi/tgsi_exec.h | 2 +-
src/gallium/drive
With that, sp_sampler_view instances are not abused anymore as a local
storage, so we can later make them constant.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 36 +---
src/gallium/drivers/softpipe/sp_tex_sample.h | 4 +---
2 files changed, 23 insertions(+), 17 dele
This is to prepare for making tgsi_sampler parameter in query_lod a
const too. These functions do not modify anything in either sampler or
view anymore.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 51 ++--
src/gallium/drivers/softpipe/sp_tex_sample.h | 4 +--
2 file
This stops abusing sp_sampler_view as a local storage and thus
allowing to put some more const modifiers in front of some types.
I might have gotten carried away too bit with the last commit. :) So
if you don't like it, then please ignore it.
Tested it with piglit shader tests and got no regressi
Those functions actually could always take them as constants.
---
src/gallium/drivers/softpipe/sp_tex_sample.c | 68 ++--
src/gallium/drivers/softpipe/sp_tex_sample.h | 4 +-
2 files changed, 36 insertions(+), 36 deletions(-)
diff --git a/src/gallium/drivers/softpipe/sp_t
On Fri, Sep 11, 2015 at 12:36 PM, 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.
>
> v1: Brian Paul: c
On Sun, Sep 06, 2015 at 06:12:38PM +0200, Francisco Jerez wrote:
> This series implements dynamic partitioning of the L3 cache space
> among its clients, the purpose is multiple:
>
> - Steal a chunk of L3 space when necessary and reserve it for SLM as
>required to support compute shaders with
On Thu, Sep 10, 2015 at 12:20 PM, Chad Versace
wrote:
> On Wed 19 Aug 2015, Anuj Phogat wrote:
> > V2:
> > - Do the tile width/height computations in the new helper
> > function and use it later in intel_miptree_get_tile_masks().
> > - Change the name to intel_get_tile_dims().
> >
> > Cc: Ben W
On Fri, Sep 11, 2015 at 10:46 AM, Hans de Goede wrote:
> Hi,
>
> I've been working on trying to fix this one:
>
> https://bugs.freedesktop.org/show_bug.cgi?id=90871
>
> And today I've more or less root caused this, it seems
> that some code is making glTexImage2D calls with npot
> width / height,
Hi,
didnt Jason just send a similar patch to the list?
[Mesa-dev] [PATCH] nir/lower_outputs_to_temporaries: Reparent the output
name
Regards
Michael
Am 11/09/2015 um 09:24 schrieb Eduardo Lima Mitev:
> Commit 1dbe4af9c9e318525fc082b542b93fb7f1e5efba
> "nir: Add a pass to lower outputs to tempor
Cc'ing stable
On Fri, Sep 11, 2015 at 8:13 AM, Alejandro Piñeiro wrote:
> opt_register_coalesce stopped to check previous instructions to
> coalesce with if somebody else was writing on the same
> destination. This can be optimized to check if somebody else was
> writing to the same channels of t
On Fri, Sep 11, 2015 at 9:00 AM, Connor Abbott wrote:
> On Fri, Sep 11, 2015 at 11:52 AM, Jason Ekstrand wrote:
>> Fortunately, nir_constant_expr already auto-splats if "dst" never shows up
>> in the constant expression field so we don't need to do anything there.
>>
>> Cc: Connor Abbott
>> ---
On Fri, Sep 11, 2015 at 11:52 AM, Jason Ekstrand wrote:
> Fortunately, nir_constant_expr already auto-splats if "dst" never shows up
> in the constant expression field so we don't need to do anything there.
>
> Cc: Connor Abbott
> ---
> src/glsl/nir/nir.h| 6 ++
> src/glsl/ni
On Fri, Sep 11, 2015 at 5:45 AM, Eduardo Lima Mitev wrote:
> Reviewed-by: Eduardo Lima Mitev
One side-note: Could you please reply-all when reviewing patches.
That way you keep the Cc list alive. While I still get it, my e-mail
client flags things that are specifically Cc'd to me so I notice th
Previously, we were passing the shader around, we were just calling it
"mem_ctx". However, the nir_shader is (and must be for the purposes of
mark-and-sweep) the mem_ctx so we might as well pass it around explicitly.
---
src/glsl/nir/nir_lower_vec_to_movs.c | 14 --
1 file changed, 8
Cc: Connor Abbott
---
src/mesa/drivers/dri/i965/brw_shader.cpp | 8
src/mesa/drivers/dri/i965/brw_vec4_nir.cpp | 6 +++---
2 files changed, 11 insertions(+), 3 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/brw_shader.cpp
b/src/mesa/drivers/dri/i965/brw_shader.cpp
index cf9aa23.
Now that we have a replicating fdot instruction, we can actually coalesce
into the destinations of vec4 instructions. We couldn't really do this
before because, if the destination had to end up in .z, we couldn't
reswizzle the instruction. With a replicated destination, the result ends
up in all
Fortunately, nir_constant_expr already auto-splats if "dst" never shows up
in the constant expression field so we don't need to do anything there.
Cc: Connor Abbott
---
src/glsl/nir/nir.h| 6 ++
src/glsl/nir/nir_opcodes.py | 3 +++
src/glsl/nir/nir_opt_algebraic.py | 3
v2 (Jason Ekstrand):
- Use nir_instr_rewrite_dest
- Pass impl directly into lower_vec_to_movs
---
src/glsl/nir/nir_lower_vec_to_movs.c | 21 +++--
1 file changed, 15 insertions(+), 6 deletions(-)
diff --git a/src/glsl/nir/nir_lower_vec_to_movs.c
b/src/glsl/nir/nir_lower_vec_to_
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.
v1: Brian Paul: code style fix.
v2: Ian Romanick: glDrawArrays(GL_QUADS, 0, (n * 4)
On Fri, Sep 11, 2015 at 1:20 AM, Connor Abbott wrote:
> On Thu, Sep 10, 2015 at 10:54 PM, Rob Clark wrote:
>>
>> On Sep 10, 2015 7:39 PM, "Jason Ekstrand" wrote:
>>>
>>> On Thu, Sep 10, 2015 at 2:39 PM, Rob Clark wrote:
>>> > From: Rob Clark
>>> >
>>> > So this is basically working as a loweri
On Thu, Sep 10, 2015 at 8:03 PM, Erik Faye-Lund wrote:
> On Thu, Sep 10, 2015 at 10:08 PM, Rob Clark wrote:
>> From: Rob Clark
>>
>> Rather than make yet another copy of channel(), let's move it into nir.
>>
>> Signed-off-by: Rob Clark
>> ---
>> src/glsl/nir/nir_builder.h | 6
On 09/10/2015 02:50 AM, Jason Ekstrand wrote:
> 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
opt_register_coalesce stopped to check previous instructions to
coalesce with if somebody else was writing on the same
destination. This can be optimized to check if somebody else was
writing to the same channels of the same destination using the
writemask.
Shader DB results (taking into account o
On 09/11/2015 07:26 AM, Roland Scheidegger wrote:
Am 11.09.2015 um 05:15 schrieb Ilia Mirkin:
My hope was (as you can see in the last patch) to enable this for all
drivers that support MS textures. I've got nv50/nvc0/r600g covered.
RadeonSI will have to read the data from the texture descriptor
Hi,
I've been working on trying to fix this one:
https://bugs.freedesktop.org/show_bug.cgi?id=90871
And today I've more or less root caused this, it seems
that some code is making glTexImage2D calls with npot
width / height, which fails on nv3x (where as it works
on nv4x).
The bug has a simple
On 09/10/2015 09:15 PM, Ilia Mirkin wrote:
This is enabled for all drivers that support MS textures. The image
component of the ext is a no-op since there is no image support in
gallium (yet).
Signed-off-by: Ilia Mirkin
---
src/mesa/state_tracker/st_extensions.c | 1 +
src/mesa/state_tra
On 09/10/2015 09:15 PM, Ilia Mirkin wrote:
Signed-off-by: Ilia Mirkin
---
src/gallium/auxiliary/tgsi/tgsi_info.c | 3 ++-
src/gallium/docs/source/tgsi.rst | 12 +++-
src/gallium/include/pipe/p_shader_tokens.h | 1 +
3 files changed, 14 insertions(+), 2 deletions(-)
2015-08-30 15:07 GMT+08:00 Boyan Ding :
> Instructions with difference in PM field can actually be paired up if
> the one without PM doesn't do packing/unpacking and non-NOP
> packing/unpacking operations from PM instruction aren't added to the
> other without PM.
>
> total instructions in shared p
On Fri, Sep 11, 2015 at 5:18 AM, Alejandro Piñeiro wrote:
> opt_register_coalesce stopped to check previous instructions to
> coalesce with if somebody else was writing on the same
> destination. This can be optimized to check if somebody else was
> writing to the same channels of the same destina
Am 11.09.2015 um 05:14 schrieb Dave Airlie:
> On 11 September 2015 at 12:37, Roland Scheidegger wrote:
>> Just on a very quick glance, seems somewhat odd this function is named
>> "cypress" whereas the other 64bit ones are named "cayman".
>> In any case, saying enabling support for chips which hav
Am 11.09.2015 um 05:15 schrieb Ilia Mirkin:
> Signed-off-by: Ilia Mirkin
> ---
> src/gallium/auxiliary/tgsi/tgsi_info.c | 3 ++-
> src/gallium/docs/source/tgsi.rst | 12 +++-
> src/gallium/include/pipe/p_shader_tokens.h | 1 +
> 3 files changed, 14 insertions(+), 2 deletio
Am 11.09.2015 um 05:15 schrieb Ilia Mirkin:
> My hope was (as you can see in the last patch) to enable this for all
> drivers that support MS textures. I've got nv50/nvc0/r600g covered.
>
> RadeonSI will have to read the data from the texture descriptor. I'm
> totally unfamiliar with LLVM, the des
Reviewed-by: Eduardo Lima Mitev
On 09/10/2015 10:58 PM, Jason Ekstrand wrote:
> We copy the output, make the old output the temporary, and give the
> temporary a new name. The copy keeps the pointer to the old name. This
> works just fine up until the point where we lower things to SSA and dele
Oops, I missed a patch from Jason from two days ago, fixing exactly the
same thing.
http://lists.freedesktop.org/archives/mesa-dev/2015-September/094107.html
Please ignore this one.
Eduardo
On 09/11/2015 09:24 AM, Eduardo Lima Mitev wrote:
> Commit 1dbe4af9c9e318525fc082b542b93fb7f1e5efba
> "ni
opt_register_coalesce stopped to check previous instructions to
coalesce with if somebody else was writing on the same
destination. This can be optimized to check if somebody else was
writing to the same channels of the same destination using the
writemask.
Shader DB results (taking into account o
Reviewed-by: Iago Toral Quiroga
On Thu, 2015-09-10 at 09:04 -0600, Brian Paul wrote:
> The sum of two unsigned ints is always >= 0. Found with Coverity.
> ---
> src/gallium/drivers/svga/svga_state_tss.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/src/gallium/driver
On Fri, 2015-09-11 at 09:24 +0200, Eduardo Lima Mitev wrote:
> Commit 1dbe4af9c9e318525fc082b542b93fb7f1e5efba
> "nir: Add a pass to lower outputs to temporary variables" messed up output
> variable names. The issue can be reproduced by dumping the NIR shaders
> with INTEL_DEBUG="vs,fs".
> ---
> s
On 11/09/15 00:35, Erik Faye-Lund wrote:
On Mon, Sep 7, 2015 at 3:54 PM, 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
Cypress/Cayman/Aruba, earlier r6xx/r7xx chips only support a subset
of the needed fp64 ops, and don't do GL4 anyway.
Signed-off-by: Glenn Kennard
---
Changes since v1:
Updated commit message
docs/GL3.txt | 4 ++--
docs/relnotes/11.1.0.html| 2 +-
src/gallium
Only for Cypress/Cayman/Aruba, older chips have only partial fp64 support.
Uses float intermediate values so only accurate for int24 range, which
matches what the blob does.
Signed-off-by: Glenn Kennard
---
Changes since v1:
Split into two functions
Make names a bit clearer which chips they app
> -Original Message-
> From: Ian Romanick [mailto:i...@freedesktop.org]
> Sent: Wednesday, September 09, 2015 8:54 PM
> To: Predut, Marius; mesa-dev@lists.freedesktop.org
> Subject: Re: [Mesa-dev] [PATCH v2] i915: fixing driver crashes if too few
> vertices are submitted
>
> On 09/09/201
> -Original Message-
> From: Eirik Byrkjeflot Anonsen [mailto:ei...@eirikba.org]
> Sent: Wednesday, September 09, 2015 9:18 PM
> To: Predut, Marius; mesa-dev@lists.freedesktop.org
> Subject: Re: [Mesa-dev] [PATCH v1] i915: fixing driver crashes if too few
> vertices are submitted
>
> Mar
On 09/10/2015 02:50 AM, Jason Ekstrand wrote:
> We made this switch in the FS backend some time ago and it seems to make a
> number of things a bit easier.
>
The commit log could be a bit less abstract, like mentioning why this
change helps the subsequent passes. In any case, if it was only for
co
Reviewed-by: Edward O'Callaghan
--
Edward O'Callaghan
edward.ocallag...@koparo.com
On Fri, Sep 11, 2015, at 02:43 PM, Dave Airlie wrote:
> From: Dave Airlie
>
> This just puts these in one place and #defines them.
>
> Signed-off-by: Dave Airlie
> ---
> src/gallium/drivers/r600/evergree
This series is,
Reviewed-by: Edward O'Callaghan
--
Edward O'Callaghan
edward.ocallag...@koparo.com
On Fri, Sep 11, 2015, at 08:09 PM, Glenn Kennard wrote:
> Series is:
>
> Reviewed-by: Glenn Kennard
> ___
> mesa-dev mailing list
> mesa-dev@list
Series is:
Reviewed-by: Glenn Kennard
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
This series is,
Reviewed-by: Edward O'Callaghan
--
Edward O'Callaghan
edward.ocallag...@koparo.com
On Fri, Sep 11, 2015, at 01:15 PM, Ilia Mirkin wrote:
> My hope was (as you can see in the last patch) to enable this for all
> drivers that support MS textures. I've got nv50/nvc0/r600g cove
Now that UBOs are uploaded as push constants. We need to obtain and
append the amount of push constant entries generated by the UBO entry
fetches to the 3DSTATE_CONSTANT_* packets.
v2: GEN8 support
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/brw_state_upload.c | 12 ++--
Use the gather table generated from the uniform uploads and
ir_binop_ubo_load to gather and pack the constants to the gather pool.
Note that the 3DSTATE_CONSTANT_* packet now refers to the gather
pool generated by the resource streamer instead of the constant buffer
pointed to by an offset of the
When the const block and offset are immediate values. Otherwise just
fall-back to the previous method of uploading the UBO constant data to
GRF using pull constants.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/brw_vec4.cpp| 13
src/mesa/drivers/dri/i965/brw_vec4
Since we now consider UBOs as push constants, we need to layout
our push constant register space in such a way that UBO registers
are packed right after uniform registers.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/brw_vec4.cpp | 38 ++
1 file ch
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 by trial and error that
disabling HW gather constant when the constant state for a stage
needs to be nullified is the only way
Switches on push constants whenever we have UBO entries.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/gen7_wm_state.c | 2 +-
src/mesa/drivers/dri/i965/gen8_ps_state.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/mesa/drivers/dri/i965/gen7_wm_state.c
If there are UBO constant entries, append them to stage_state->push_const_size.
The gather pool contains the combined entries of both ordinary uniforms
and UBO constants.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/gen6_vs_state.c | 20 ++--
1 file changed, 18 in
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/brw_vec4.cpp | 12
src/mesa/drivers/dri/i965/brw_vec4.h | 1 +
2 files changed, 13 insertions(+)
diff --git a/src/mesa/drivers/dri/i965/brw_vec4.cpp
b/src/mesa/drivers/dri/i965/brw_vec4.cpp
index 85dc372..f2b03f8 10064
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 track of these constant data
fetches into a gather table.
The gather table is generated from two sources. Ordinary uniform fet
This patch generates the gather table entries for ordinary uniforms
if they are present. The uniform constants here will later be packed
together with UBO constants.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/brw_fs.cpp | 18 +-
1 file changed, 17 insertions(+),
We now have two sources of constant buffers: UBOs and ordinary uniforms.
After assigning a block of push constant hw-register to normal uniforms,
just pack the UBO push constant registers right after it.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/brw_fs.cpp | 20 ---
When the const block and offset are immediate values. Otherwise just
fall-back to the previous method of uploading the UBO constant data to
GRF using pull constants.
Cc: kenn...@whitecape.org
Cc: ja...@jlekstrand.net
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/brw_fs.cpp
Now that we consider UBO constants as push constants, we need to include
the sizes of the UBO's constant slots in the visitor's uniform slot sizes.
This information is needed to properly pack vector constants tightly next to
each other.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i96
Reserve space in the gather pool where the gathered uniforms are flushed.
Signed-off-by: Abdiel Janulgue
---
src/mesa/drivers/dri/i965/gen6_vs_state.c | 8
1 file changed, 8 insertions(+)
diff --git a/src/mesa/drivers/dri/i965/gen6_vs_state.c
b/src/mesa/drivers/dri/i965/gen6_vs_state.
To be able to refer to a constant buffer, the resource streamer needs
to index it with a hardware binding table entry. This blankets the ubo
buffers with hardware binding table indices.
Gather constants hardware fetches in 16-entry binding table blocks.
So we need to use a block that is unused.
S
Signed-off-by: Abdiel Janulgue
---
src/glsl/nir/nir_types.cpp | 6 ++
src/glsl/nir/nir_types.h | 2 ++
2 files changed, 8 insertions(+)
diff --git a/src/glsl/nir/nir_types.cpp b/src/glsl/nir/nir_types.cpp
index 940c676..a0caf0e 100644
--- a/src/glsl/nir/nir_types.cpp
+++ b/src/glsl/nir/nir
1 - 100 of 113 matches
Mail list logo
