Reviewed-by: Bas Nieuwenhuizen
for the series.
On Wed, Sep 12, 2018 at 11:19 PM Samuel Pitoiset
wrote:
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/si_cmd_buffer.c | 14 ++
> 1 file changed, 6 insertions(+), 8 deletions(-)
>
> diff -
radv_queue *queue,
> uint32_t scratch_size,
> @@ -2244,6 +2271,20 @@ radv_get_preamble_cs(struct radv_queue *queue,
> if (scratch_bo)
> radv_cs_add_buffer(queue->de
Reviewed-by: Bas Nieuwenhuizen
On Thu, Sep 13, 2018 at 4:36 PM Samuel Pitoiset
wrote:
>
> Same code is generated because LLVM ends up by using bfe, but
> that seems cleaner to me.
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_nir_to_llvm.c | 21 ++
Reviewed-by: Bas Nieuwenhuizen
for the series
On Wed, Sep 12, 2018 at 3:40 PM Samuel Pitoiset
wrote:
>
> Indirect descriptors only need one entry, we don't have to
> emit a location for every descriptors.
>
> Fixes GPU hangs with new CTS:
> dEQP-VK.binding_model.descriptors
> R_028C4C_PA_SC_CONSERVATIVE_RASTERIZATION_CNTL,
> + pa_sc_conservative_rast |
> + S_028C4C_PREZ_AA_MASK_ENABLE(1) |
> + S_028C4C_POSTZ_AA_MASK_ENABLE(1) |
> +
On Mon, Sep 10, 2018 at 8:05 PM Jason Ekstrand wrote:
>
> The instruction scheduler is re-ordering loads which is causing fence
> values to be loaded after the value they're fencing. In particular,
> consider the following pseudocode:
>
> void try_use_a_thing(int idx)
> {
> bool
On Mon, Sep 10, 2018 at 8:05 PM Jason Ekstrand wrote:
>
> ---
> src/compiler/spirv/spirv_to_nir.c | 170 ++
> 1 file changed, 103 insertions(+), 67 deletions(-)
>
> diff --git a/src/compiler/spirv/spirv_to_nir.c
> b/src/compiler/spirv/spirv_to_nir.c
> index
though.
>
> On Mon, Sep 10, 2018 at 1:35 PM Bas Nieuwenhuizen
> wrote:
>>
>> ---
>> src/amd/vulkan/radv_device.c | 7 +++
>> src/amd/vulkan/radv_extensions.py | 2 +-
>> 2 files changed, 8 insertions(+), 1 deletion(-)
>>
>> diff --git a/src/amd/vul
I tried, but I can't find from your driver what the HW does, so for
this review the same assumption that the CTS tests pass for you.
Reviewed-by: Bas Nieuwenhuizen
On Mon, Sep 10, 2018 at 7:08 PM Jason Ekstrand wrote:
>
> Cc: Bas Nieuwenhuizen
> ---
> src/intel/vulkan/anv_devic
---
src/amd/vulkan/radv_device.c | 7 +++
src/amd/vulkan/radv_extensions.py | 2 +-
2 files changed, 8 insertions(+), 1 deletion(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index 53f99a8cecd..7917ed7ffe5 100644
--- a/src/amd/vulkan/radv_device.c
+++
Acked-by: Bas Nieuwenhuizen
On Mon, Sep 10, 2018 at 7:09 PM Jason Ekstrand wrote:
>
> ---
> include/vulkan/vulkan_core.h | 139 +++--
> src/vulkan/registry/vk.xml | 280 +++
> 2 files changed, 345 insertions(+), 74 deletions(-)
On Tue, Sep 4, 2018 at 6:03 PM Dylan Baker wrote:
>
> Quoting Bas Nieuwenhuizen (2018-09-02 17:35:49)
> > Just mirror the radeonsi bits. Since this is just adding the extra
> > switch entries for new HW I think this should be fine for stable.
> >
> > CC
Mirrors
1e40f694831 "ac/surface: fix CMASK fast clear for NPOT textures with mipmapping
on SI/CI/VI"
CC:
---
src/amd/vulkan/radv_image.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/amd/vulkan/radv_image.c b/src/amd/vulkan/radv_image.c
index
Just mirror the radeonsi bits. Since this is just adding the extra
switch entries for new HW I think this should be fine for stable.
CC:
---
src/amd/vulkan/radv_pipeline.c | 1 +
src/amd/vulkan/si_cmd_buffer.c | 1 +
2 files changed, 2 insertions(+)
diff --git a/src/amd/vulkan/radv_pipeline.c
No clue what gets fixed by this but both radeonsi and amdvlk do it.
CC:
---
src/amd/vulkan/radv_device.c | 24 ++--
1 file changed, 22 insertions(+), 2 deletions(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index c300c88468a..3d208ab053d
0, Emil Velikov wrote:
> >>>>
> >>>>
> >>>> Hi Timothy,
> >>>>
> >>>> On 31 August 2018 at 10:57, Timothy Arceri
> >>>> wrote:
> >>>>>
> >>>>>
&
+TImothy
On Fri, Aug 31, 2018 at 11:32 AM Alexander Larsson wrote:
>
> Hi, I'm the developer behind flatpak (https://flatpak.org/) and we've
> recently run into some issues with the mesa shader cache. Flatpak has
> a app/runtime split where the runtime is shared between app and
> provides /usr.
Reviewed-by: Bas Nieuwenhuizen
On Fri, Aug 31, 2018 at 7:56 AM Dave Airlie wrote:
>
> From: Dave Airlie
>
> ac_surface.c: gfx6_compute_surface says
> /* DB doesn't support linear layouts. */
>
> Now if we expose linear depth and create a linear depth image
> and
okay, r-b with that.
On Thu, Aug 30, 2018 at 1:39 PM Samuel Pitoiset
wrote:
>
>
>
> On 8/30/18 12:30 PM, Bas Nieuwenhuizen wrote:
> > On Wed, Aug 29, 2018 at 10:55 PM Samuel Pitoiset
> > wrote:
> >>
> >> CTS doesn't test input clip/cull distances
On Wed, Aug 29, 2018 at 10:55 PM Samuel Pitoiset
wrote:
>
> CTS doesn't test input clip/cull distances for the fragment
> shader stage, which explains why this was totally broken. I
> wrote a simple test locally that works now.
>
> This fixes a crash with GTA V and DXVK.
>
> Note that we are
gt; > store->src[2] = nir_src_for_ssa(nir_ssa_undef(, 1, 32));
> > @@ -557,6 +559,7 @@ build_nir_itoi_compute_shader(struct radv_device *dev,
> > bool is_3d)
> >
> > nir_ssa_def *outval = >dest.ssa;
> > nir_intrinsic_instr *store = nir_intrinsic_instr_create(
CC:
---
src/amd/vulkan/radv_formats.c | 19 +++
1 file changed, 19 insertions(+)
diff --git a/src/amd/vulkan/radv_formats.c b/src/amd/vulkan/radv_formats.c
index f0cc0fc5f95..6253c27b95d 100644
--- a/src/amd/vulkan/radv_formats.c
+++ b/src/amd/vulkan/radv_formats.c
@@ -1112,6
Reviewed-by: Bas Nieuwenhuizen
On Wed, Aug 29, 2018 at 5:53 AM Dave Airlie wrote:
>
> From: Dave Airlie
>
> While adding transfer queues to radv, I started writing some tests,
> the first test I wrote fell over copying a buffer larger than this
> limit.
>
> Checked AMDV
Reviewed-by: Bas Nieuwenhuizen
On Tue, Aug 28, 2018 at 8:41 PM Marek Olšák wrote:
>
> From: Marek Olšák
>
> This fixes VM faults and corruption.
>
> Cc: 18.1 18.2
> ---
> src/amd/common/ac_surface.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
Reviewed-by: Bas Nieuwenhuizen
On Tue, Aug 21, 2018 at 9:49 AM, Samuel Pitoiset
wrote:
> This should be applied on top of "ac: add imad & fmad helpers".
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_nir_to_llvm.c | 24
&g
On Tue, Aug 21, 2018 at 12:38 AM, Marek Olšák wrote:
> On Fri, Aug 10, 2018 at 9:26 AM Rob Clark wrote:
>>
>> Used internally in freedreno/ir3 for the vec2 value that hw passes to
>> shader to use as coordinate for bary.f (varying fetch) instruction.
>> This is not the same as
On Mon, Aug 20, 2018 at 11:32 PM, Grazvydas Ignotas wrote:
> Thanks to reproducible builds, binary file timestamps may be identical
> for both 32bit and 64bit packages when built from the same source.
> This means radv will use the same cache for both 32 and 64 bit
> processes, which leads to
rent GPUs ...
For now, this series is
Reviewed-by: Bas Nieuwenhuizen
> ---
> src/amd/vulkan/radv_meta.c | 16 +++-
> 1 file changed, 7 insertions(+), 9 deletions(-)
>
> diff --git a/src/amd/vulkan/radv_meta.c b/src/amd/vulkan/radv_meta.c
> index b8d0a7cfc70..1ec8896afa2 10
Reviewed-by: Bas Nieuwenhuizen
Did you have access to push?
On Mon, Aug 20, 2018 at 11:40 PM, Grazvydas Ignotas wrote:
> For 32bit build. Trivial.
> ---
> src/vulkan/wsi/wsi_common_display.c | 4 ++--
> src/vulkan/wsi/wsi_common_x11.c | 2 +-
> 2 files changed, 3 i
Reviewed-by: Bas Nieuwenhuizen
On Fri, Aug 17, 2018 at 10:06 PM, Jason Ekstrand wrote:
> Found by inspection. This doesn't help much now but we'll see this
> pattern with images if you load UNORM and then store UNORM.
>
> Shader-db results on Kaby Lake:
>
> total inst
('ishl', 'a@16', 16), 16), ('extract_i16', a, 0),
> '!options->lower_extract_word'),
> + (('ishr', 'a@16', 16), ('extract_i16', a, 1),
> '!options->lower_extract_word'),
> (('iand', 0x, a), ('extract_u16', a, 0),
> '!options->lower_extract_word')
On Fri, Aug 17, 2018 at 10:06 PM, Jason Ekstrand wrote:
> If it's not the right bit-size, it may not actually be the correct
> extraction. For now, we'll only worry about 32-bit versions.
>
> Fixes: 905ff8619824 "nir: Recognize open-coded extract_u16"
> Fixes: 76289fbfa84a "nir: Recognize
Reviewed-by: Bas Nieuwenhuizen
for this patch.
On Fri, Aug 17, 2018 at 4:32 PM, Kai Wasserbäch
wrote:
> Supresses a maybe-uninitialized warning with GCC 8.
>
> Note: image_index should always be initialised due to the result check,
> but the compiler doesn't see that.
Thanks!
Reviewed-by: Bas Nieuwenhuizen
On Fri, Aug 17, 2018 at 2:45 PM, Samuel Pitoiset
wrote:
> When the number of unique BO is 0, we optimize the list creation
> by copying all buffers of the current CS directly into it. But
> this is only valid if the CS doesn't have virtua
Reviewed-by: Bas Nieuwenhuizen
On Thu, Aug 16, 2018 at 12:33 PM, Samuel Pitoiset
wrote:
> We have to do a fast-clear eliminate when clearing DCC
> metadata with 0x20202020. I don't know if that fixes anything
> but that seems correct to me.
>
> CC: 18.2
> Signed-off-b
Reviewed-by: Bas Nieuwenhuizen
On Thu, Aug 16, 2018 at 11:22 AM, Samuel Pitoiset
wrote:
> This was missing when VK_EXT_conditional_rendering has been
> implemented. The predication type should be -1 to avoid
> restoring previous state when performing a decompression pass
> with
On Wed, Aug 15, 2018 at 6:08 PM, Dylan Baker wrote:
> Quoting Bas Nieuwenhuizen (2018-08-09 17:27:46)
>> CC:
>> ---
>> src/amd/vulkan/Android.mk | 2 ++
>> src/amd/vulkan/Makefile.am | 2 +-
>> 2 files changed, 3 insertions(+), 1 deletion(-)
>>
>>
Seems like DXVK depends on that and it might get reverted
upstream. Since apps are not supposed to use 0 in v2 anyway,
we should be safe implementing the old behavior there.
Fixes: 66e12451ac4 "radv: Update to new VK_EXT_vertex_attribute_divisor to
version 2."
CC: 18.2
---
On Wed, Aug 15, 2018 at 11:49 AM, Jason Ekstrand wrote:
> On Tue, Aug 14, 2018 at 7:20 AM Samuel Pitoiset
> wrote:
>>
>> Reviewed-by: Samuel Pitoiset
>>
>> On 7/23/18 4:24 PM, Bas Nieuwenhuizen wrote:
>> > Behavior wrt firstInstance got changed, and
Seems that in a single case we use the renderpass before checking
the pipeline, so check the renderpass before we use it.
Fixes: fbcd1673144 "radv: Add on-demand compilation of built-in shaders."
---
src/amd/vulkan/radv_meta_resolve_fs.c | 8
1 file changed, 8 insertions(+)
diff --git
Reviewed-by: Bas Nieuwenhuizen
On Wed, Aug 15, 2018 at 3:28 PM, Samuel Pitoiset
wrote:
> Reported by Coverity.
>
> Fixes: fbcd167314 ("radv: Add on-demand compilation of built-in shaders.")
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_meta.c |
On Wed, Aug 15, 2018 at 2:13 PM, Samuel Pitoiset
wrote:
>
>
> On 8/14/18 9:15 PM, Bas Nieuwenhuizen wrote:
>>
>> On Tue, Aug 14, 2018 at 6:11 PM, Samuel Pitoiset
>> wrote:
>>>
>>> The last parameter of radeon_set_sh_reg_seq() is the number
Reviewed-by: Bas Nieuwenhuizen
On Wed, Aug 15, 2018 at 3:09 PM, Samuel Pitoiset
wrote:
> This option allows us to remove additional s_waitcnt instructions
> because s_barrier internally does s_waitcnt 0.
>
> Though, apparently there is a problem with LDS accesses that
> causes r
On Wed, Aug 15, 2018 at 2:04 PM, Mauro Rossi wrote:
> Hi Bas,
>
> Il giorno mar 14 ago 2018 alle ore 23:10 Bas Nieuwenhuizen
> ha scritto:
>>
>> On Tue, Aug 14, 2018 at 10:48 PM, Mauro Rossi
>> wrote:
>> > (VkShaderModule) cast is added before NULL
On Wed, Aug 15, 2018 at 1:16 PM, Mike Lothian wrote:
> Do you need the game name here, isn't that set in driconf?
There is no driconf for vulkan/radv yet.
Reviewed-by: Bas Nieuwenhuizen
I assume you tested that it works with the workaround?
>
> On Wed, 15 Aug 2018 at 12:12 Timot
On Tue, Aug 14, 2018 at 10:48 PM, Mauro Rossi wrote:
> (VkShaderModule) cast is added before NULL to avoid following building error:
>
> external/mesa/src/amd/vulkan/radv_meta_decompress.c:365:54: error:
> incompatible pointer to integer conversion passing 'void *' to parameter
> of type
compute pipelines, so we don't need to initialize it.
Note that there we don't set WAVES_PER_SH. Is that intended?
Reviewed-by: Bas Nieuwenhuizen
Also a radeonsi patch?
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/si_cmd_buffer.c | 4 +---
> 1 file changed, 1 inse
---
src/amd/vulkan/radv_meta_blit.c | 525 +++-
1 file changed, 175 insertions(+), 350 deletions(-)
diff --git a/src/amd/vulkan/radv_meta_blit.c b/src/amd/vulkan/radv_meta_blit.c
index f68496d3202..370d4fcd831 100644
--- a/src/amd/vulkan/radv_meta_blit.c
+++
While at it, share the exemplars and account for a non-occurring
fs key.
---
src/amd/vulkan/radv_meta_blit.c | 21 +++--
src/amd/vulkan/radv_meta_blit2d.c | 19 ++-
src/amd/vulkan/radv_meta_clear.c | 19 ++-
In environments where we cannot cache, e.g. Android (no homedir),
ChromeOS (readonly rootfs) or sandboxes (cannot open cache), the
startup cost of creating a device in radv is rather high, due
to compiling all possible built-in pipelines up front. This meant
depending on the CPU a 1-4 sec cost of
On Mon, Aug 13, 2018 at 5:54 PM, Dylan Baker wrote:
> Quoting Emil Velikov (2018-08-13 03:38:13)
>> On 10 August 2018 at 00:57, Dylan Baker wrote:
>> > Quoting Chad Versace (2018-08-09 10:37:33)
>> >> On Tue 07 Aug 2018, Dylan Baker wrote:
>> >> > Q
CC:
---
src/amd/vulkan/Android.mk | 2 ++
src/amd/vulkan/Makefile.am | 2 +-
2 files changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/amd/vulkan/Android.mk b/src/amd/vulkan/Android.mk
index cee3744f40b..51b03561fa7 100644
--- a/src/amd/vulkan/Android.mk
+++ b/src/amd/vulkan/Android.mk
On Thu, Aug 9, 2018 at 7:48 PM, Chad Versace wrote:
> On Wed 08 Aug 2018, Bas Nieuwenhuizen wrote:
>> This became kind of messy as python imports cannot really look up
>> parent/sibling directories. I saw some scripts use sys.path but
>> that became even more messy du
On Thu, Aug 9, 2018 at 7:49 PM, Chad Versace wrote:
> On Wed 08 Aug 2018, Bas Nieuwenhuizen wrote:
>> ---
>> src/vulkan/Makefile.am| 3 +
>> src/vulkan/util/meson.build | 2 +
>> src/vulkan/util/vk_entrypoints_gen.py | 515 ++
s Legg.
This fixes random CTS fails with
dEQP-VK.renderpass.suballocation.formats.d32_sfloat_s8_uint.input.*
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105396
CC:
Signed-off-by: Samuel Pitoiset
Reviewed-by: Bas Nieuwenhuizen
>
> On Wed, 2018-03-28 at
On Wed, Aug 8, 2018 at 12:39 PM, Eric Engestrom
wrote:
> On Wednesday, 2018-08-08 11:31:09 +0100, Emil Velikov wrote:
>> On 8 August 2018 at 00:14, Bas Nieuwenhuizen
>> wrote:
>> > This became kind of messy as python imports cannot really look up
>> > paren
On Wed, Aug 8, 2018 at 12:19 PM, Emil Velikov wrote:
> On 8 August 2018 at 00:14, Bas Nieuwenhuizen wrote:
>
>> @@ -146,7 +148,7 @@ static const uint16_t string_map[${strmap.hash_size}] = {
>> };
>>
>> int
>> -${name_prefix}_string_map_looku
This became kind of messy as python imports cannot really look up
parent/sibling directories. I saw some scripts use sys.path but
that became even more messy due to import locations.
I also move the selections of the dispatch table out of the
generation script because it is not easily shared, and
radv does not need them and the trampolines need a dispatch
table in the instance and device.
---
src/vulkan/util/vk_entrypoints_gen.py | 47 +++
1 file changed, 27 insertions(+), 20 deletions(-)
diff --git a/src/vulkan/util/vk_entrypoints_gen.py
messy stuff in the build system due to
difficulties with python includes. I tested with meson and
autotools. Android.mk is updated but not tested.
Bas Nieuwenhuizen (4):
vulkan: Add central copy of entrypoints/extensions code.
vulkan/util: Add support to not generate the trampolines.
anv
---
src/vulkan/Makefile.am| 3 +
src/vulkan/util/meson.build | 2 +
src/vulkan/util/vk_entrypoints_gen.py | 515 ++
src/vulkan/util/vk_extensions.py | 92 +
src/vulkan/util/vk_extensions_gen.py | 205 ++
5 files changed, 817
Replace some functions to use the anv idioms. However, I did not
introduce dispatch tables in the instance/device since that seemed
like overkill.
---
src/amd/vulkan/Android.mk | 16 +-
src/amd/vulkan/Makefile.am | 29 +-
src/amd/vulkan/meson.build | 91
On Fri, Jul 27, 2018 at 11:40 AM, Timothy Arceri wrote:
>
>
> On 27/07/18 19:10, Bas Nieuwenhuizen wrote:
>>
>> On Fri, Jul 27, 2018 at 7:32 AM, Timothy Arceri
>> wrote:
>>>
>>> Because this was setting image to true we would end up calling
>
Follow radeonsi.
Fixes: 3665f66ef26 "radv: Add support for ETC2 textures."
---
src/amd/vulkan/radv_formats.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/amd/vulkan/radv_formats.c b/src/amd/vulkan/radv_formats.c
index f23ebfb2ad7..f0cc0fc5f95 100644
---
Did this pass CTS? If so, woot!
Reviewed-by: Bas Nieuwenhuizen
btw we can avoid some more, I think literally everything on Vega uses
L2 so the VERTEX_ATTRIBUTE_READ, TRANSFER_READ (this one could use CP
DMA, but not for images, and CP DMA can be configured to go through
L2
Reviewed-by: Bas Nieuwenhuizen
for the series
On Fri, Jul 27, 2018 at 11:28 AM, Samuel Pitoiset
wrote:
> This wasn't wrong but it looks better to me like this. It's
> only used for debugging purposes (ie. RADV_TRACE_FILE).
>
> Signed-off-by: Samuel Pitoiset
> ---
&g
Reviewed-by: Bas Nieuwenhuizeh
for the series.
On Fri, Jul 27, 2018 at 11:55 AM, Samuel Pitoiset
wrote:
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_cmd_buffer.c | 31 +++
> 1 file changed, 23 insertions(+), 8 deletions(-)
>
> diff --git
Reviewed-by: Bas Nieuwenhuizen
Does it occur often that they are different?
On Fri, Jul 27, 2018 at 5:36 AM, Marek Olšák wrote:
> From: Marek Olšák
>
> a needle in the haystack?
>
> Cc: 18.1
> ---
> src/amd/common/ac_surface.c | 2 +-
> 1 file changed, 1 insertion(+)
On Fri, Jul 27, 2018 at 7:32 AM, Timothy Arceri wrote:
> Because this was setting image to true we would end up calling
> si_load_image_desc() when we sould be calling
> si_load_sampler_desc().
Since the descriptor is part of an image, not a sampler,
get_image_descriptor looks like the right
Reviewed-by: Bas Nieuwenhuizen
Fixes: 506a07e4e3a "ac/nir: Add deref support to image intrinsics."
On Fri, Jul 27, 2018 at 7:08 AM, Timothy Arceri wrote:
> Looks like a mistake from when the deref stuff landed.
>
> Fixes: 506a07e4e3a4 ("ac/nir: Add deref support to im
On Thu, Jul 26, 2018 at 5:55 PM, Emil Velikov wrote:
> On 26 July 2018 at 16:25, Bas Nieuwenhuizen wrote:
>> On Thu, Jul 26, 2018 at 5:15 PM, Bas Nieuwenhuizen
>> wrote:
>>> On Thu, Jul 26, 2018 at 5:01 PM, Emil Velikov
>>> wrote:
>>>>
On Thu, Jul 26, 2018 at 5:15 PM, Bas Nieuwenhuizen
wrote:
> On Thu, Jul 26, 2018 at 5:01 PM, Emil Velikov
> wrote:
>> Hi Bas,
>>
>> Pardon for joining so late.
>>
>> On 18 July 2018 at 13:01, Bas Nieuwenhuizen wrote:
>>> radv always needs it,
On Thu, Jul 26, 2018 at 5:01 PM, Emil Velikov wrote:
> Hi Bas,
>
> Pardon for joining so late.
>
> On 18 July 2018 at 13:01, Bas Nieuwenhuizen wrote:
>> radv always needs it, so just check the header instead. Also
>> do not declare the function if the variable is n
On Wed, Jul 25, 2018 at 3:38 PM, Samuel Pitoiset
wrote:
>
>
> On 07/25/2018 03:28 PM, Bas Nieuwenhuizen wrote:
>>
>> On Wed, Jul 25, 2018 at 2:30 PM, Samuel Pitoiset
>> wrote:
>>>
>>> This breaks printing input/output variables with more than
>>
On Wed, Jul 25, 2018 at 3:28 PM, Samuel Pitoiset
wrote:
>
>
> On 07/25/2018 03:26 PM, Bas Nieuwenhuizen wrote:
>>
>> On Wed, Jul 25, 2018 at 3:25 PM, Samuel Pitoiset
>> wrote:
>>>
>>>
>>>
>>> On 07/25/2018 02:55 PM, Bas Nieuwenh
On Wed, Jul 25, 2018 at 2:30 PM, Samuel Pitoiset
wrote:
> This breaks printing input/output variables with more than
> 4 components like mat4.
>
> Fixes: 1beef89ad8 ("nir: prepare for bumping up max components to 16")
> Signed-off-by: Samuel Pitoiset
> ---
> src/compiler/nir/nir_print.c | 1 -
>
On Wed, Jul 25, 2018 at 3:25 PM, Samuel Pitoiset
wrote:
>
>
> On 07/25/2018 02:55 PM, Bas Nieuwenhuizen wrote:
>>
>> hmm, not sure why that did not fail on my CTS runs ...
>
>
> Because 16bit is only enabled on Polaris?
Wait what, it should be enabled on Vega
hmm, not sure why that did not fail on my CTS runs ...
Reviewed-by: Bas Nieuwenhuizen
On Wed, Jul 25, 2018 at 2:55 PM, Samuel Pitoiset
wrote:
> Move the integer conversion after the fixup.
>
> This fixes some regressions with
> dEQP-VK.pipeline.vertex_input.single_attribute.mat4.as
That we don't have a background disk cache does not mean we should
prevent the app caching anything.
CC:
---
src/amd/vulkan/radv_pipeline_cache.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/src/amd/vulkan/radv_pipeline_cache.c
b/src/amd/vulkan/radv_pipeline_cache.c
On Thu, Jul 19, 2018 at 2:48 PM, Daniel Schürmann
wrote:
> Signed-off-by: Daniel Schürmann
> ---
> src/compiler/glsl_types.h | 15 +++
> src/compiler/nir_types.cpp | 12
> src/compiler/nir_types.h | 1 +
> 3 files changed, 28 insertions(+)
>
> diff --git
Behavior wrt firstInstance got changed, and a divisor of 0 has been
disallowed.
The new version of the ext got published in specification 1.1.81.
---
src/amd/vulkan/radv_extensions.py | 2 +-
src/amd/vulkan/radv_nir_to_llvm.c | 7 ---
2 files changed, 5 insertions(+), 4 deletions(-)
diff
On Mon, Jul 23, 2018 at 1:50 PM, Samuel Pitoiset
wrote:
> When we keep shader info for eg. RADV_TRACE_FILE, we don't
> free the NIR when creating the pipelines. Though, we have to
> free it when destroying the shader module.
>
> Signed-off-by: Samuel Pitoiset
> ---
>
Patches 1-2 are
Reviewed-by: Bas Nieuwenhuizen
Patches 3-4 are
Acked-by: Bas Nieuwenhuizen
but I'm not that knowledgeable about the Android build system.
On Sun, Jul 22, 2018 at 10:14 AM, Mauro Rossi wrote:
> The following series adds support for vulkan radv for Android 8.1
>
> [
Patches 1-2 are
Reviewed-by: Bas Nieuwenhuizen
On Sun, Jul 22, 2018 at 10:14 AM, Mauro Rossi wrote:
> Patch changes radv entrypoints generator to not skip this extension even
> though it is set as disabled in the vk.xml
>
> Reference: 63525ba730 ("android: enable VK_ANDR
On Sat, Jul 21, 2018 at 11:32 AM, Mauro Rossi wrote:
> This patch applies the necessary changes in Android.common.mk
> as per automake rules, to avoid following building error:
>
> external/mesa/src/gallium/drivers/nouveau/nouveau_screen.c:159:8:
> error: implicit declaration of function
ugh, yet another buildsystem.
Reviewed-by: Bas Nieuwenhuizen
On Sat, Jul 21, 2018 at 11:32 AM, Mauro Rossi wrote:
> This patch applies the necessary changes in Android.common.mk
> as per automake rules, to avoid following building error:
>
> external/mesa/src/gallium/dri
CC stable?
Reviewed-by: Bas Nieuwenhuizen
On Fri, Jul 20, 2018 at 6:48 PM, Samuel Pitoiset
wrote:
> modules[i] can be NULL for merged shaders but we have to
> free the NIR code. radv_can_dump_shader_stats() already handles
> if modules[i] is NULL, no need to check it twice.
>
Reviewed-by: Bas Nieuwenhuizen
Thanks!
On Fri, Jul 20, 2018 at 6:47 PM, Samuel Pitoiset
wrote:
> Users shouldn't use this debugging option except when we
> ask them to do!
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_device.c | 4
> 1 file
There always is a continue block, so let us just do unreachable.
Fixes: 8cacf38f527 "nir: Do not use continue block after removing it."
CC: 18.1
CC: Mark Janes
---
src/compiler/nir/nir_opt_if.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/src/compiler/nir/nir_opt_if.c
Reviewed-by: Bas Nieuwenhuizen
On Fri, Jul 20, 2018 at 3:07 PM, Samuel Pitoiset
wrote:
> That shouldn't be needed because the DB state is invalid.
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_meta_resolve_fs.c | 3 +--
> 1 file changed, 1 inserti
Reviewed-by: Bas Nieuwenhuizen
for the series.
On Wed, Jul 18, 2018 at 4:19 PM, Samuel Pitoiset
wrote:
> The goal is to use radv_barrier()/radv_subpass_barrier() as
> much as possible for further optimizations.
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan
On Thu, Jul 19, 2018 at 6:48 PM, Eric Engestrom
wrote:
> On Wednesday, 2018-07-18 14:01:49 +0200, Bas Nieuwenhuizen wrote:
>> radv always needs it, so just check the header instead. Also
>> do not declare the function if the variable is not set, so we
>> get a nice
radv always needs it, so just check the header instead. Also
do not declare the function if the variable is not set, so we
get a nice compile error instead of failing to open a device
at runtime.
Fixes: b87ef9e606a "util: fix MSVC build issue in disk_cache.h"
---
configure.ac | 1 +
Reviewed-by: Bas Nieuwenhuizen
Thanks for the improvements!
On Wed, Jul 18, 2018 at 10:47 AM, Samuel Pitoiset
wrote:
> Inherited commands buffers are not supported.
>
> v3: - restore previous conditional rendering user state for fast clears
> v2: - disable predication for b
Reviewed-by: Bas Nieuwenhuizen
On Tue, Jul 17, 2018 at 5:03 PM, Samuel Pitoiset
wrote:
> We don't need to emit PS_PARTIAL_FLUSH for the pre fragment shader
> stages (ie. geometry/tessellation). Emitting VS_PARTIAL_FLUSH
> is enough for these stages. Note that PS_PARTIAL_F
Reviewed-by: Bas Nieuwenhuizen
On Fri, Jul 13, 2018 at 5:35 PM, Samuel Pitoiset
wrote:
> If we know that the given image doesn't have any metadata,
> we don't need to flush.
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_cmd_buffer.c | 20 ++--
Reviewed-by: Bas Nieuwenhuizen
On Mon, Jul 16, 2018 at 9:20 PM, Timothy Pearson
wrote:
>
> color_interp_vgpr_index was declared as a generic char value.
> Because signed values are used in this variable, the result
> was not safe across architectures and crashed on ppc64[e
Aligning with AMDVLK on GFX9. I don't think it changes much.
On Tue, Jul 17, 2018 at 9:15 AM, Samuel Pitoiset
wrote:
> Why do we need that? Are you sure it's correct?
>
>
> On 07/14/2018 02:28 PM, Bas Nieuwenhuizen wrote:
>>
>> When no occlusion queries are active even if
Reviewed-by: Bas Nieuwenhuizen
On Thu, Jul 12, 2018 at 4:26 PM, Samuel Pitoiset
wrote:
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/winsys/amdgpu/radv_amdgpu_cs.c | 8 ++--
> 1 file changed, 2 insertions(+), 6 deletions(-)
>
> diff --git a/src/amd/vu
Reviewed-by: Bas Nieuwenhuizen
On Fri, Jul 13, 2018 at 2:14 PM, Samuel Pitoiset
wrote:
> The Vulkan 1.1.80 spec says:
>
> "viewMask has the same effect for the described subpass as
> VkRenderPassMultiviewCreateInfo::pViewMasks has on each
> corresponding subpass."
501 - 600 of 2350 matches
Mail list logo