Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Mon, Apr 23, 2018 at 2:17 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> ---
> src/amd/common/ac_gpu_info.h | 16
&
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Mon, Apr 23, 2018 at 2:43 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This refactors the code out to share it between radv and radeonsi.
> ---
&
On Mon, Apr 23, 2018 at 2:20 AM, Timothy Arceri <tarc...@itsqueeze.com> wrote:
>
>
> On 23/04/18 03:26, Bas Nieuwenhuizen wrote:
>>
>> We seems to use progress for two cases:
>
>
> seems -> seem
>
>> 1) When we lowered some returns.
>> 2)
We seems to use progress for two cases:
1) When we lowered some returns.
2) When we remove unreachable code.
If just case 2 happens we assert as state->return_flag has not
been allocated yet, but we are still trying to do insert all
predicates based on it.
This splits the concerns. We only use
On Fri, Apr 20, 2018 at 5:16 PM, Jason Ekstrand wrote:
> On Fri, Apr 20, 2018 at 5:16 AM, Nicolai Hähnle wrote:
>>
>> On 20.04.2018 10:21, Iago Toral wrote:
>>>
>>> Hi,
>>>
>>> while developing support for Vulkan shaderInt16 on Anvil I came across
>>> a
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Fri, Apr 20, 2018 at 2:21 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> The SI family doesn't support chaining which means the maximum
> size in dwords per CS is limited. When that limit was reached
>
Otherwise a lot of games complain about not having enough memory,
and it is sort of local so this seems reasonable to me.
CC: 18.0
---
src/amd/vulkan/radv_device.c | 8 +---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git
No clue about travis, but from RADV side
Acked-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
Thanks!
On Thu, Apr 19, 2018 at 4:18 PM, Juan A. Suarez Romero
<jasua...@igalia.com> wrote:
> This is a backport for 18.0 from 6ce400782c ("travis: radeonsi and radv
> nee
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Tue, Apr 17, 2018 at 4:05 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> When DCC is enabled with MSAA textures, CMASK should be
> cleared to 0x.
>
> Signed-off-by: Samue
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Wed, Apr 18, 2018 at 6:53 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Might be useful for debugging purposes, especially when we
> want to replace a shader on the fly.
>
> Signed-off-by: Samue
I have seen a few applications and games do the dynamic buffer bounds
incorrectly, this
make it easier to work around, e.g. for debugging.
---
src/amd/vulkan/radv_cmd_buffer.c | 4 +++-
src/amd/vulkan/radv_debug.h | 1 +
src/amd/vulkan/radv_device.c | 1 +
3 files changed, 5
---
src/amd/vulkan/radv_device.c | 5 -
src/amd/vulkan/radv_pipeline.c | 3 ++-
src/amd/vulkan/radv_shader.c | 1 +
src/amd/vulkan/si_cmd_buffer.c | 4
4 files changed, 11 insertions(+), 2 deletions(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index
On Wed, Apr 18, 2018 at 2:11 PM, Marek Olšák <mar...@gmail.com> wrote:
> On Wed, Apr 18, 2018 at 4:44 PM, Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
> wrote:
>>
>> IIRC if level N is unaligned then num_dcc_levels <= N+1, so level N+1
>> is not DCC compress
IIRC if level N is unaligned then num_dcc_levels <= N+1, so level N+1
is not DCC compressed?
On Wed, Apr 18, 2018 at 12:53 PM, Marek Olšák wrote:
> On Wed, Apr 18, 2018 at 5:54 AM, Nicolai Hähnle wrote:
>>
>> On 17.04.2018 02:41, Marek Olšák wrote:
>>>
>>>
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Apr 17, 2018 at 3:08 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> This fixes some random CTS failures:
>
> dEQP-VK.renderpass.multisample.*.
>
> Performing a fast-clear eliminate is still usel
On Mon, Apr 16, 2018 at 1:17 PM, Juan A. Suarez Romero
<jasua...@igalia.com> wrote:
> On Mon, 2018-04-16 at 00:09 +0200, Bas Nieuwenhuizen wrote:
>> No clue how I missed those ...
>>
>> Fixes: 4503ff760c "ac/nir: Add workaround for GFX9 buffer views."
>
Reviewed-by: Bas Niuwenhuizen
for the series. Thanks for the cleanup.
On Fri, Apr 13, 2018 at 7:14 PM, Samuel Pitoiset
wrote:
> When decompressing DCC we don't enable it, so it's useless
> to disable it. This reduces the number of prediction
No clue how I missed those ...
Fixes: 4503ff760c "ac/nir: Add workaround for GFX9 buffer views."
CC:
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105320
---
src/amd/common/ac_nir_to_llvm.c | 39 +++--
1 file changed, 22
On Fri, Apr 13, 2018 at 8:06 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
>
>
> On 04/12/2018 01:44 AM, Bas Nieuwenhuizen wrote:
>>
>> ---
>> src/compiler/shader_info.h| 1 +
>> src/compiler/spirv/spirv_to_nir.c | 6 ++
>> 2 fi
Okay, this series is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
Please split out the patch Jason commented on and give me a link to a
branch and I'll merge.
On Tue, Apr 10, 2018 at 4:37 PM, Daniel Schürmann
<daniel.schuerm...@campus.tu-berlin.de> wrote:
> ---
&g
Test test to the list.
On Thu, Apr 12, 2018 at 9:28 AM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Looks good to me, do you have tests?
>
> Reviewed-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
>
>
> On 04/12/2018 01:45 AM, Bas Nieuwenhuizen wrote:
>&
From: Bas Nieuwenhuizen <ba...@chromium.org>
---
Makefile.am | 2 +
.../vertex_attribute_divisor.c| 216 ++
2 files changed, 218 insertions(+)
create mode 100644
src/tests/func/vertex_attribute_d
On Thu, Apr 12, 2018 at 1:20 PM, Samuel Pitoiset
wrote:
> Both fixed resolve and FS resolve methods don't support integer
> formats, even if destination image is DCC compressed.
>
> No CTS changes on Vega, but this helps fixing some tests
> when DCC is enabled for MSAA
This adds everything except non-uniform indexing, which needs a bit
more work and testing.
---
src/amd/vulkan/radv_device.c | 39 +++
src/amd/vulkan/radv_extensions.py | 1 +
src/amd/vulkan/radv_shader.c | 2 ++
3 files changed, 42 insertions(+)
diff --git
Pretty straight forward, just pass the divisors through the shader
key and then do a LLVM divide.
---
src/amd/vulkan/radv_device.c | 6 ++
src/amd/vulkan/radv_extensions.py | 1 +
src/amd/vulkan/radv_nir_to_llvm.c | 26 +++---
src/amd/vulkan/radv_pipeline.c| 26
---
src/compiler/shader_info.h| 1 +
src/compiler/spirv/spirv_to_nir.c | 6 ++
2 files changed, 7 insertions(+)
diff --git a/src/compiler/shader_info.h b/src/compiler/shader_info.h
index ababe520b2d..c8128fea01b 100644
--- a/src/compiler/shader_info.h
+++ b/src/compiler/shader_info.h
The continue means we do alignment differently than during creation,
making the buffer smaller than expected.
---
src/amd/vulkan/radv_descriptor_set.c | 3 ---
1 file changed, 3 deletions(-)
diff --git a/src/amd/vulkan/radv_descriptor_set.c
b/src/amd/vulkan/radv_descriptor_set.c
index
---
src/compiler/shader_info.h| 1 +
src/compiler/spirv/spirv_to_nir.c | 4
2 files changed, 5 insertions(+)
diff --git a/src/compiler/shader_info.h b/src/compiler/shader_info.h
index c8128fea01b..4a0a843c796 100644
--- a/src/compiler/shader_info.h
+++ b/src/compiler/shader_info.h
---
src/amd/vulkan/radv_descriptor_set.c | 30 +++-
src/amd/vulkan/radv_descriptor_set.h | 1 +
2 files changed, 30 insertions(+), 1 deletion(-)
diff --git a/src/amd/vulkan/radv_descriptor_set.c
b/src/amd/vulkan/radv_descriptor_set.c
index 7a3a611dd68..9b35451c497
---
src/amd/vulkan/radv_descriptor_set.c | 21 +
1 file changed, 17 insertions(+), 4 deletions(-)
diff --git a/src/amd/vulkan/radv_descriptor_set.c
b/src/amd/vulkan/radv_descriptor_set.c
index 9b35451c497..55b4aaa388c 100644
--- a/src/amd/vulkan/radv_descriptor_set.c
+++
---
src/compiler/spirv/spirv.core.grammar.json | 169 -
src/compiler/spirv/spirv.h | 18 +++
2 files changed, 183 insertions(+), 4 deletions(-)
diff --git a/src/compiler/spirv/spirv.core.grammar.json
b/src/compiler/spirv/spirv.core.grammar.json
index
---
src/amd/vulkan/radv_cmd_buffer.c | 4 --
src/amd/vulkan/radv_debug.c | 3 -
src/amd/vulkan/radv_descriptor_set.c | 82 +---
src/amd/vulkan/radv_descriptor_set.h | 4 --
src/amd/vulkan/radv_private.h| 2 -
5 files changed, 13 insertions(+), 82
Previously we did not care about havin the set storage in order,
but for variable descriptor count we want the highest binding
at the end of the storage.
---
src/amd/vulkan/radv_descriptor_set.c | 43 ++--
1 file changed, 41 insertions(+), 2 deletions(-)
diff --git
With update after bind we can't attach bo's to the command buffer
from the descriptor set anymore, so we have to have a global BO
list.
I am somewhat surprised this works really well even though we have
implicit synchronization in the WSI based on the bo list associations
and with the new
This adds support for VK_EXT_descriptor_indexing except for the
non-uniform indexing, which should be sent shortly.
Please review!
Bas Nieuwenhuizen (10):
spirv: Update spirv.h to 12f8de9f04327336b699b1b80aa390ae7f9ddbf4
radv: Keep a global BO list for VkMemory.
radv: Don't store buffer
his is the latest
> release for 17.3 series.
Can we please get this in 17.3? This has been submitted upstream
laready for 2 weeks and this backport has been available for a week
and fixes one of the major games from Feral on Vega:
commit 4503ff760c794c3bb15b978a47c530037d564
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Wed, Apr 11, 2018 at 2:09 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> And add some comments.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
>
On Wed, Apr 4, 2018 at 10:19 PM, Bas Nieuwenhuizen
<b...@basnieuwenhuizen.nl> wrote:
> On GFX9 whether the buffer size is interpreted as elements or bytes
> depends on whether IDXEN is enabled in the instruction. If the index
> is a constant zero, LLVM optimizes IDXEN to 0.
&
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Wed, Apr 11, 2018 at 9:29 AM, Tapani Pälli <tapani.pa...@intel.com> wrote:
> Fixes linking errors against:
>
>anv_GetPhysicalDeviceImageFormatProperties2KHR
>radv_GetPhysicalDeviceImageFormatPrope
For dcn1 && < 64 bpp displayable surfaces, addrlib only accepts
S swizzles.
At the same time addrlib prefers D swizzles is allowed, so we can
just allow S swizzles as fallback.
Fixes: b64b712558 "ac/surface/gfx9: request desired micro tile mode explicitly"
---
src/amd/common/ac_surface.c | 7
What is the addrlib assertion we are hitting?
On Tue, Apr 10, 2018 at 11:44 AM, Michel Dänzer wrote:
> On 2018-04-06 07:12 PM, Marek Olšák wrote:
>> From: Marek Olšák
>>
>> This enables the tile swizzle for some cases of the displayable micro mode,
>>
---
src/amd/vulkan/radv_device.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index 4fc7392e65e..22e8f1e7a78 100644
--- a/src/amd/vulkan/radv_device.c
+++ b/src/amd/vulkan/radv_device.c
@@ -293,7 +293,8 @@
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Apr 10, 2018 at 4:00 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> The source and destination image parameters were swapped.
>
> No CTS changes on Polaris10, but I suspect this might
>
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Apr 10, 2018 at 2:09 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Otherwise, the shader BOs are not added to the list on SI because
> prefetching isn't supported. Calling radv_cs_add_buffer() in the
&
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
I'm wondering though, doesn't this result in more saves/restores, as
we now do it for each part of a subpass clear separately?
On Mon, Apr 9, 2018 at 11:10 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> This remov
Saves about 2% of compile time for F1 2017, as well as reduce code
size of an optimized libvulkan_radeon.so by about 1 KiB.
This still keeps the hashtable, as we also stored blocks in there.
---
src/amd/common/ac_nir_to_llvm.c | 22 +-
1 file changed, 13 insertions(+), 9
...@gmail.com> wrote:
>>>
>>> On Sun, Apr 8, 2018 at 5:54 PM, Bas Nieuwenhuizen
>>> <b...@basnieuwenhuizen.nl> wrote:
>>> > On Sun, Apr 8, 2018 at 11:40 PM, Rob Clark <robdcl...@gmail.com> wrote:
>>> >> On Sun, Apr 8, 20
As we sometimes reset them to -1, -1 does not mean that they are
not written by the secondary command buffer.
Fixes: ad11fc3571 "radv: don't emit unneeded vertex state."
---
src/amd/vulkan/radv_cmd_buffer.c | 17 +++--
1 file changed, 3 insertions(+), 14 deletions(-)
diff --git
The packet can sometimes be skipped, but we still think the change takes effect.
This just makes the packet always take effect.
Fixes: ad11fc3571 "radv: don't emit unneeded vertex state."
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105942
---
src/amd/vulkan/radv_cmd_buffer.c | 2 +-
Okay, this needs more work, looks like almost everything got reset
somewhere else already, and num_instances wasn't but should not be
needed.
On Mon, Apr 9, 2018 at 4:37 PM, Bas Nieuwenhuizen
<b...@basnieuwenhuizen.nl> wrote:
> Fixes: ad11fc3571 "radv: don't emit unneede
Fixes: ad11fc3571 "radv: don't emit unneeded vertex state."
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=105942
---
src/amd/vulkan/radv_cmd_buffer.c | 5 +
1 file changed, 5 insertions(+)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Apr 9, 2018 at 2:38 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Forgot one check... Too many mistakes for a simple change.
>
> Fixes: f1d7c16e85 ("radv: fix prefetching compute shaders
According to Marek, not enabling it on Stoney has a significant
negative performance impact. (And I guess this might impact
performance on Raven as well)
The register settings are pretty much copied from radeonsi. I did
not put this in the pipeline as that would make the pipeline more
dependent
On Mon, Apr 9, 2018 at 10:12 AM, Samuel Pitoiset
wrote:
> Simple extension that only returns information for AMD hw.
>
> v2: - update computation of computeUnitsPerShaderArray
>
> Signed-off-by: Samuel Pitoiset
> ---
>
On Sun, Apr 8, 2018 at 11:40 PM, Rob Clark <robdcl...@gmail.com> wrote:
> On Sun, Apr 8, 2018 at 5:20 PM, Bas Nieuwenhuizen
> <b...@basnieuwenhuizen.nl> wrote:
>>>>>>>>>> +
>>>>>>>>>> + /** The mode of the underlying
On Sun, Apr 8, 2018 at 10:43 PM, Bas Nieuwenhuizen
<b...@basnieuwenhuizen.nl> wrote:
> On Sun, Apr 8, 2018 at 10:23 PM, Bas Nieuwenhuizen
> <b...@basnieuwenhuizen.nl> wrote:
>> On Tue, Apr 3, 2018 at 8:32 PM, Jason Ekstrand <ja...@jlekstrand.net> wrote:
>>>
+
+ /** The mode of the underlying variable */
+ nir_variable_mode mode;
>>>
>>> In fact, it seems like deref->mode is unused outside of nir_print and
>>> nir_validate.. for logical addressing we can get the mode from the
>>> deref_var->var at the
On Sun, Apr 8, 2018 at 10:23 PM, Bas Nieuwenhuizen
<b...@basnieuwenhuizen.nl> wrote:
> On Tue, Apr 3, 2018 at 8:32 PM, Jason Ekstrand <ja...@jlekstrand.net> wrote:
>> This commit adds a new instruction type to NIR for handling derefs.
>> Nothing uses it yet but this adds
On Sun, Apr 8, 2018 at 6:06 PM, Rob Clark <robdcl...@gmail.com> wrote:
> On Sun, Apr 8, 2018 at 11:15 AM, Bas Nieuwenhuizen
> <b...@basnieuwenhuizen.nl> wrote:
>> On Sun, Apr 8, 2018 at 3:29 PM, Rob Clark <robdcl...@gmail.com> wrote:
>>> On Sun, Apr 8, 20
On Tue, Apr 3, 2018 at 8:32 PM, Jason Ekstrand wrote:
> This commit adds a new instruction type to NIR for handling derefs.
> Nothing uses it yet but this adds the data structure as well as all of
> the code to validate, print, clone, and [de]serialize them.
> ---
>
On Sun, Apr 8, 2018 at 3:29 PM, Rob Clark <robdcl...@gmail.com> wrote:
> On Sun, Apr 8, 2018 at 8:58 AM, Bas Nieuwenhuizen
> <b...@basnieuwenhuizen.nl> wrote:
>> On Sun, Apr 8, 2018 at 1:38 PM, Rob Clark <robdcl...@gmail.com> wrote:
>>> On Tue, Apr
On Sun, Apr 8, 2018 at 1:38 PM, Rob Clark wrote:
> On Tue, Apr 3, 2018 at 2:32 PM, Jason Ekstrand wrote:
>> This commit adds a new instruction type to NIR for handling derefs.
>> Nothing uses it yet but this adds the data structure as well as all of
>>
On Fri, Apr 6, 2018 at 2:28 PM, Samuel Pitoiset
wrote:
> Simple extension that only returns information for AMD hw.
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_device.c | 71
> +++
Thanks. The series is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Fri, Apr 6, 2018 at 7:34 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Hi,
>
> This small series is a preliminary work before doing some
> improvements in the DCC/CMASK/FMAS
According to Marek, not enabling it on Stoney has a significant
negative performance impact. (And I guess this might impact
performance on Raven as well)
The register settings are pretty much copied from radeonsi. I did
not put this in the pipeline as that would make the pipeline more
dependent
On Thu, Apr 5, 2018 at 11:42 AM, Samuel Pitoiset
wrote:
> ... to radv_flush_vertex_buffers().
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_cmd_buffer.c | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Apr 5, 2018 at 10:27 AM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Enable it directly in the preamble, but do not enable line
> on Polaris10/11/12 because there is a hw bug.
>
> There is pos
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Apr 5, 2018 at 10:27 AM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> This unnecessary when the precision bit flag is not set, and this
> might hurt performance. The Vulkan explain
On Thu, Apr 5, 2018 at 10:32 AM, Samuel Pitoiset
wrote:
> Ported from RadeonSI.
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_image.c | 5 +
> 1 file changed, 5 insertions(+)
>
> diff --git a/src/amd/vulkan/radv_image.c
On GFX9 whether the buffer size is interpreted as elements or bytes
depends on whether IDXEN is enabled in the instruction. If the index
is a constant zero, LLVM optimizes IDXEN to 0.
Now the size in elements is interpreted in bytes which of course
results in out of bounds accesses.
The correct
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Wed, Apr 4, 2018 at 12:12 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> This allows to start draws as soon as possible.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com&
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Wed, Apr 4, 2018 at 10:55 AM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Ported from RadeonSI.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
> s
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Mar 29, 2018 at 2:51 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> For GFX9+ only, RadeonSI does this too.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
> sr
Has someone grown tests for this since the time I sent patches for
enabling this to the list?
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Apr 2, 2018 at 6:17 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> The driver already supports expo
Series is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Mar 29, 2018 at 3:24 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Some will be used for further optimizations (ie. out-of-order rast).
>
> Signed-off-by: Samuel Pitoiset <s
00, James Legg wrote:
>> On Thu, 2018-03-22 at 02:36 +0100, Bas Nieuwenhuizen wrote:
>> > On Thu, Mar 8, 2018 at 12:59 PM, James Legg <jl...@feralinteractive.com>
>> > wrote:
>> > > This avoids bug 105396 somehow. I suspect it is a VI and GFX9 hardware
>&
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Sun, Mar 25, 2018 at 8:15 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> The driver only supports the required formats for now.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gma
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Mar 20, 2018 at 10:07 AM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> It's a 32-bit integer like the layer.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
> src/
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, 27 Mar 2018, 10:08 Samuel Pitoiset, <samuel.pitoi...@gmail.com>
wrote:
> Tested-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
>
> On 03/27/2018 02:39 AM, Marek Olšák wrote:
> > From:
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Thu, Mar 22, 2018 at 4:41 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Based on RadeonSI. Untested.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
> s
I'd rather we figure out a story of when it is faster. I tried a lot
of stuff with the currently available games, and getting it
consistently faster was difficult.
So if you have a raven ridge, feel free to try something, but I'm not
really a fan of copying it blindly without understanding it.
isl: Don't use surface format R32_FLOAT for typed atomic integer
> operations
> intel/compiler: Memory fence commit must always be enabled for gen10+
>
> Bas Nieuwenhuizen (6):
> radv: Always lower indirect derefs after nir_lower_global_vars_to_local.
> vul
ists.freedesktop.org>
> CC: Dave Airlie <airl...@redhat.com>
> CC: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
> CC: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
> src/amd/vulkan/radv_cmd_buffer.c | 52
> +---
>
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Wed, Mar 21, 2018 at 9:30 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> The hardware only supports 32-bit depth surfaces, but we can
> enable TC-compat HTILE for 16-bit depth surfa
On Tue, Mar 20, 2018 at 3:11 PM, Samuel Pitoiset
wrote:
> Ported from AMVDVLK.
>
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_pipeline.c | 14 ++
> 1 file changed, 14 insertions(+)
>
> diff --git
is compiler/nir/nir.h the preferred way?
I checked src/amd and we are using "nir/nir.h" all the time.
On Tue, Mar 20, 2018 at 12:41 PM, Emil Velikov wrote:
> From: Emil Velikov
>
> Stay consistent with the rest of the codebase, effectively
The reason they cannot be enabled at the same time is because the
behavior was different. I see nothing about using the AMD behavior in
this patch? Also since IIRC maintenance1 is core in vulkan 1.1 we
cannot reasonably expose the AMD ext if the instance was created with
vulkan >= 1.1
On Mon, Mar
I think we still need to support image->image copies with depth
outside the range [0,1]. There was no explicit restriction on that,
but before enabling this it is impossible to get an image with depth
not in [0,1] so we did not have to care. But the gfx path uses the
depth HW with no depth bound
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 19, 2018 at 4:42 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This fixes:
> dEQP-VK.multiview.input_attachments*
> ---
> src/amd/vulkan/radv_shad
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 19, 2018 at 5:30 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> If a shader only writes to an output via a constant initializer we
> n
FWIW
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Thu, Mar 15, 2018 at 3:27 PM, Rob Clark <robdcl...@gmail.com> wrote:
> Since a fair bit of access to nir_deref_var's is just to chase down the
> nir_variable for an intrinsic (or texture/sa
One comment on patch 7, and the LLVM dump ordering issue we debugged
earlier (where we should dump before compiling), otherwise the series
is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Wed, Mar 14, 2018 at 12:27 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wro
On Wed, Mar 14, 2018 at 12:27 PM, Samuel Pitoiset
wrote:
> Signed-off-by: Samuel Pitoiset
> ---
> src/amd/vulkan/radv_shader.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/src/amd/vulkan/radv_shader.c
On Thu, Mar 15, 2018 at 6:33 AM, Jason Ekstrand wrote:
> This commit adds a new instruction type to NIR for handling derefs.
> Nothing uses it yet but this adds the data structure as well as all of
> the code to validate, print, clone, and [de]serialize them.
>
> Cc: Rob
On Thu, Mar 15, 2018 at 10:46 AM, Daniel Schürmann
<daniel.schuerm...@campus.tu-berlin.de> wrote:
> With AMD_gcn_shader renamed to gcn_shader, this patch is
>
> Reviewed-by: Daniel Schürmann
Note that the mailaddress is mangled a bit.
Reviewed-by: Bas Nieuwenhuizen <b...@b
Thanks a lot!
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Sat, Mar 10, 2018 at 7:42 PM, Jason Ekstrand <ja...@jlekstrand.net> wrote:
> On Sat, Mar 10, 2018 at 10:18 AM, Jason Ekstrand <ja...@jlekstrand.net>
> wrote:
>>
>> This
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
for the series.
On Tue, Mar 13, 2018 at 3:06 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
> src/amd/common/ac_nir_to_llvm.c |
Thanks, 1-8 are
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
Patch 9 is
Acked-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 12, 2018 at 3:52 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
>
>
> On 03/12/2018 03:25 PM, Samuel Pitois
On Mon, Mar 12, 2018 at 2:50 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
>
>
> On 03/12/2018 02:48 PM, Bas Nieuwenhuizen wrote:
>>
>> On Mon, Mar 12, 2018 at 2:48 PM, Samuel Pitoiset
>> <samuel.pitoi...@gmail.com> wrote:
>>>
>>>
On Mon, Mar 12, 2018 at 2:48 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
>
>
> On 03/12/2018 02:24 PM, Bas Nieuwenhuizen wrote:
>>
>> Hi Samuel,
>>
>> Can we put the code into a separate file, instead of into radv_shader.c?
>
>
> If
801 - 900 of 2350 matches
Mail list logo