To catch more of those hangs early.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cs.h | 4
1 file changed, 4 insertions(+)
diff --git a/src/amd/vulkan/radv_cs.h b/src/amd/vulkan/radv_cs.h
index 2c8935f3065..8fcdad9dbbd 100644
--- a/src/amd/vulkan/rad
On Mon, Mar 27, 2017 at 10:14 PM, Dave Airlie wrote:
> From: Dave Airlie
>
> If we had no rasterization, we'd emit SPI color
> format as all 0's the hw dislikes this, add the workaround
> from radeonsi.
>
> Found while debugging tessellation
>
>
It reads @ writes the DB cache, and we haven't flushed dst caches yet,
so DB cache may be stale. Also the user might be shader read (and probably is),
so also flush after.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
CC: <mesa-sta...@lists.freedesktop.org>
Fixes: f4e499ec791
v_pipeline_has_gs(cmd_buffer->state.pipeline))
> primgroup_size = 64; /* recommended with a GS */
>
> multi_instances_smaller_than_primgroup = indirect_draw ||
> (instanced_draw &&
> num_prims
> <
x_se;
I just realized we can alloc 16 KiB instead of 32 KiB for HAWAII
(based on tess_offchip_block_dw_size).
Either way:
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
> - hs_offchip_param = radv_get_hs_offchip_param(queue->device,
> -
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Sun, Mar 19, 2017 at 5:18 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This was meant to be checking the index type to get the correct
> index not the last e
patch 1-4 are
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
and should be in my opinion pushable before any of the discussed
changes to the disk cache.
On Wed, Mar 15, 2017 at 5:17 AM, Timothy Arceri <tarc...@itsqueeze.com> wrote:
> Apps can limit the size o
Thanks. Pushed.
On Tue, Mar 14, 2017 at 4:26 PM, Alex Smith wrote:
> Need to flush before updating the buffer to ensure that the copy is
> ordered after previous accesses (assuming the app has performed the
> appropriate barriers).
>
> This fixes potential issues due
again. Looking at that site, the -pro
driver also uses VK_MAKE_VERSION, so keeping consistency is probably
best.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_device.c | 24 +++-
1 file changed, 23 insertions(+), 1 deletion(-)
diff --git a/s
I've skimmed to changes from 1.0.5 to 1.0.42 and I think we have all
changes. We're still not conformant ofcourse, but this should not
regress stuff,
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_device.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
The flushes could be due to TRANSFER barriers.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
Cc: 17.0 <mesa-sta...@lists.freedesktop.org>
---
src/amd/vulkan/si_cmd_buffer.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/amd/vulkan/si_cmd_buffer.c b/sr
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 20, 2017 at 12:03 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> The current code evaluated to always true, we only want to flush
> on the first submit. R
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Fri, Mar 17, 2017 at 5:24 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This just uses an 8-bit clear and packs the values.
>
> Signed-off-by: Dave Airlie <
Pushed.
On Thu, Mar 16, 2017 at 5:40 PM, Alex Smith wrote:
> The index passed to get_shared_memory_ptr is an attribute slot index,
> i.e. the index of a vec4 within LDS. Therefore this must be scaled by
> sizeof(vec4) to give the LDS byte offset.
>
> Signed-off-by:
Where do you see that they are required to flush denormals to 0? I
can't find it in the GLSL extensions for SPIR-V, and the vulkan spec
has the following note:
"Any denormalized value input into a shader or potentially generated
by any instruction in a shader may be flushed to 0."
Which very
We never supported more, as we can run out of user SGPRs on the VS.
Fixes: f4e499ec791 ("radv: add initial non-conformant radv vulkan driver")
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
CC: <mesa-sta...@lists.freedesktop.org>
CC: James Legg <jl...@feralinte
Pushed, but see my new patch for caveats.
On Thu, Mar 16, 2017 at 6:48 PM, James Legg wrote:
> Avoid a buffer overflow in ac_nir_to_llvm.c's create_function when
> using more than 4 descriptor sets. radv claims support for 8.
>
> Cc: 17.0
On Wed, Mar 15, 2017 at 1:25 AM, Dave Airlie wrote:
> From: Dave Airlie
>
> This adds support for exporting 2D images, to an
> opaque fd.
>
> This implements the:
> VK_KHX_external_memory_capabilities
> VK_KHX_external_memory
> VK_KHX_external_memory_fd
>
>
On Fri, Mar 17, 2017 at 1:04 AM, Dave Airlie wrote:
> From: Dave Airlie
>
> In order to get isinf(NaN) correct, at least radv can't
> use an unordered equals which feq has to be for us, this
Why do we have to use an unordered equal normally? SPIR-V has
t seems like a safer option for now.
Yeah we should be able to. I'm more comfortable sending this patch to
stable though, so this patch is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
> ---
> src/amd/vulkan/radv_cmd_buffer.c | 2 +-
> 1 file changed, 1 insertion
This reverts commit cce43f6d8c40222099badaf52344d6a0eed993f3.
Redundant, as the flush already happens at si_cp_dma_prepare.
Cc: Dave Airlie
---
src/amd/vulkan/si_cmd_buffer.c | 3 ---
1 file changed, 3 deletions(-)
diff --git a/src/amd/vulkan/si_cmd_buffer.c
I thought rand() was not thread safe anyway, and hence we can't use it
in mesa as we don't know what application threads call it too?
On Tue, Mar 14, 2017 at 3:08 AM, Timothy Arceri wrote:
> Otherwise we will always remove old cache entries from the same dirs.
> ---
>
On Wed, Mar 15, 2017 at 8:46 AM, Alex Smith <asm...@feralinteractive.com> wrote:
> On 14 March 2017 at 20:56, Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
> wrote:
>>
>> The flushes could be due to TRANSFER barriers.
>>
>> Signed-off-by: Bas Nieuwenhuize
ries is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
Did you have push access?
On Tue, Apr 4, 2017 at 11:29 PM, Fredrik Höglund <fred...@kde.org> wrote:
> All offsets and strides are precomputed by
> radv_CreateDescriptorUpdateTemplateKHR and stored in the te
The buffer sizes are specified just a few lines earlier, so don't
repeat ourselves.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_query.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/amd/vulkan/radv_query.c b/src/amd/
So I think we need to reset both command buffer states (the enable and
the index) when we call a secondary command buffer.
With that fixed, this patch is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Apr 11, 2017 at 3:30 PM, Alex Smith <asm...@feralinteractive.c
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Apr 3, 2017 at 7:17 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This fixes an old bug that seems to get triggered by
> dEQP-VK.memory.requirements.imag
We supported more generally. Decreased the dynamic buffers though, as
we only support 16 for uniform+storage.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_device.c | 28 ++--
1 file changed, 14 insertions(+), 14 deletions(-)
diff
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Fri, Mar 31, 2017 at 10:58 PM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> Fixes: 2b35b60d: radv: move to using nir clip/cull merge pass.
>
> Signed-off-by
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Sun, Apr 2, 2017 at 6:47 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> y is vert, x is horiz.
>
> Noticed in visual inspection compared to radeonsi.
>
On Thu, Mar 30, 2017 at 10:00 AM, Dave Airlie wrote:
> From: Dave Airlie
>
> Signed-off-by: Dave Airlie
> ---
> src/amd/common/ac_nir_to_llvm.c | 4
> 1 file changed, 4 insertions(+)
>
> diff --git
On Thu, Mar 30, 2017 at 10:00 AM, Dave Airlie wrote:
> From: Dave Airlie
>
> This patch adds support for the offchip rings for storing
> tessellation factors and attribute data.
>
> It includes the register setup for the TF ring
>
> Signed-off-by: Dave
viewport and takes the minimum.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/si_cmd_buffer.c | 52 ++
1 file changed, 12 insertions(+), 40 deletions(-)
diff --git a/src/amd/vulkan/si_cmd_buffer.c b/src/amd/vulkan/si_cmd_buffer.c
On Wed, Apr 12, 2017 at 12:48 AM, Dave Airlie <airl...@gmail.com> wrote:
> On 12 April 2017 at 08:19, Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl> wrote:
>> On Wed, Apr 12, 2017 at 12:04 AM, Thomas Hindoe Paaboel Andersen
>> <pho...@gmail.com> wrote:
>>&
The outer result was referred to, which meant bugs.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_query.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/amd/vulkan/radv_query.c b/src/amd/vulkan/radv_query.c
index bd293
On Wed, Apr 12, 2017 at 12:04 AM, Thomas Hindoe Paaboel Andersen
wrote:
> ---
> src/amd/vulkan/radv_wsi.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/src/amd/vulkan/radv_wsi.c b/src/amd/vulkan/radv_wsi.c
> index b8999f4..37cb322 100644
> ---
Makes more sense when we hash the layout for the pipeline cache.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/common/ac_nir_to_llvm.c | 12 ++-
src/amd/vulkan/radv_descriptor_set.c | 42
src/amd/vulkan/radv_descriptor
Since the shader code can include them.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_descriptor_set.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/amd/vulkan/radv_descriptor_set.c
b/src/amd/vulkan/radv_descriptor_set.c
index ba5d5eb75e5..c048a
Patches 1 & 6 are
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Wed, Apr 12, 2017 at 12:03 AM, Thomas Hindoe Paaboel Andersen
<pho...@gmail.com> wrote:
> From 1811ccf1
> ---
> src/amd/vulkan/radv_device.c | 2 --
> 1 file changed, 2 deletions(-)
>
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
Fixes: 8475a14302e ("radv: Implement pipeline statistics queries.")
---
src/amd/vulkan/radv_query.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/src/amd/vulkan/radv_query.c b/src/amd/vulkan/ra
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
Fixes: 8475a14302e ("radv: Implement pipeline statistics queries.")
---
src/amd/vulkan/radv_query.c | 6 ++
1 file changed, 6 insertions(+)
diff --git a/src/amd/vulkan/radv_query.c b/src/amd/vulkan/radv_query.c
i
buffer.
>
> Signed-off-by: Alex Smith <asm...@feralinteractive.com>
> Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
> ---
> src/amd/vulkan/radv_cmd_buffer.c | 55
>
> src/amd/vulkan/radv_private.h| 1 +
>
)
On Mon, Apr 10, 2017 at 4:54 AM, Edward O'Callaghan
<funfunc...@folklore1984.net> wrote:
> One trivial comment but otherwise 1&2 are,
> Reviewed-by: Edward O'Callaghan <funfunc...@folklore1984.net>
>
> On 04/10/2017 09:34 AM, Bas Nieuwenhuizen wrote:
>> Adds a
do that we may need to look a bit more
into this. Having 8 WAIT_REG_MEM packets per query doesn't sound ideal.
This also restricts the availability word in the pool to timestamp queries
only, as occlusion queries don't use it, and pipeline statistic queries
likely won't either.
Signed-off-by: Bas
Adds a shader for writing occlusion query results to a buffer, as the
CP packet isn't support on SI or secondary buffers, and doesn't handle
the availability bit (or partial results) nor truncation to 32-bit.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_query.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/src/amd/vulkan/radv_query.c b/src/amd/vulkan/radv_query.c
index cfe16a9d0e2..dc1844adb51 100644
--- a/src/amd/vulkan/radv_query.c
+++
The devil is in the shader again, otherwise this is
fairly straightforward.
The CTS contains no pipeline statistics copy to buffer
testcases, so I did a basic smoketest.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_device.c | 2 +-
src/amd/vulkan/radv_pri
For using them with both occlusion and pipeline statistics queries.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_private.h | 4 ++--
src/amd/vulkan/radv_query.c | 22 +++---
2 files changed, 13 insertions(+), 13 deletions(-)
diff --git a/s
mber of
samples, but that hasn't happened yet, so we only change it to
enable/disable counting)
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_device.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_meta_buffer.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/amd/vulkan/radv_meta_buffer.c
b/src/amd/vulkan/radv_meta_buffer.c
index 1e94f3b5866..cfa0b9320e2 100644
--- a/src/amd/vulkan/radv_meta_bu
Most trace points happen after an operation, so add a trace point
at the start of the command buffer.
Furthermore, add one after a CmdUpdateBuffer using CP_DMA as that
didn't emit one yet.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 1 +
s
This is only relevant with 0 attachments. In that case we do nothing
on subpass switch already, and the pipeline is the authoritative
source of the number of samples, so this shouldn't change anything.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_device.c | 2
Properly and with comments this time.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_device.c | 44 +---
1 file changed, 29 insertions(+), 15 deletions(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/
Needed if we want to allow them taking more than 64 KiB. The calculations
of these already used 32 bits.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_descriptor_set.h | 16
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/s
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Apr 13, 2017 at 9:29 PM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This bumps it to the same level as amdgpu-pro, it also
> moves a bunch of dEQP-VK.geometry.i
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Apr 13, 2017 at 9:35 PM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This enables a bunch of NotSupported CTS tests.
>
> Signed-off-by: Dave Airlie <
For consistency, doesn't really impact performance.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index 958ae6e361e..40e6e
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Apr 13, 2017 at 6:37 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> The addrlib import meant we'd return after we attempted
> to setup the no stencil bit
v2: Also convert the calculations.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_descriptor_set.c | 4 ++--
src/amd/vulkan/radv_private.h| 10 +-
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/src/amd/vulkan/radv_descriptor_se
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_private.h | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/src/amd/vulkan/radv_private.h b/src/amd/vulkan/radv_private.h
index 00190e7eee0..2afc0cbedfc 100644
--- a/src/amd/
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Thu, Apr 13, 2017 at 6:14 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This is ported from radeonsi, and avoids the bug in the
> addrlib code. This should probab
Gives me approximately a 2% perf increase in bot dota2 & talos.
Having descriptors (both sets and vertex buffers) prefetched
didn't help so I didn't include that.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 6 ++
src/amd/vulkan/radv
On Tue, Apr 18, 2017 at 5:57 AM, Dave Airlie wrote:
> From: Dave Airlie
>
> In practice this will probably just drop draw id in a few places.
>
> Signed-off-by: Dave Airlie
> ---
> src/amd/common/ac_nir_to_llvm.c | 42
Patch 1, 3-9 are
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Apr 18, 2017 at 5:57 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> There is some radv specific info we need to gather from shaders
> bef
Signed-off-by: Bas Nieeuwenhuizen
---
src/amd/vulkan/radv_pipeline_cache.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/amd/vulkan/radv_pipeline_cache.c
b/src/amd/vulkan/radv_pipeline_cache.c
index 5f6355f0d1a..0ab4d2a26e3 100644
---
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Sun, Apr 23, 2017 at 9:22 PM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This just updates this to use the same flags as radeonsi
> for consistency.
>
&
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index 958ae6e361e..ffa7e430b2b 100644
--- a/src/amd/
Moves from accurate divides to rcp and rsq. No new CTS failures.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/common/ac_nir_to_llvm.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
Results in faster code than the lowering by LLVM.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_pipeline.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/amd/vulkan/radv_pipeline.c b/src/amd/vulkan/radv_pipeline.c
index 8e71d59fae7..5046c9f6b36
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 7 +--
1 file changed, 1 insertion(+), 6 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index 5d4236ca187..274495f134f 100644
--- a/src/amd/
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 12 +---
1 file changed, 9 insertions(+), 3 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index 8c812084141..a1127130a5d 100644
--- a/src/amd/
Without storaes, the only writes are fast clears, transfers and metadata
initialization, each of which have the appropiate invalidations already.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 28 +++-
1 file chang
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 36 +++-
1 file changed, 19 insertions(+), 17 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index 5b44ce84529..07d0a
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 16 +++-
1 file changed, 11 insertions(+), 5 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index 07d0a0c2c12..088a3c9025c 100644
--- a/s
The only way we write CMASK/DCC compressed textures through shaders
is fast clears and CMASK/DCC inits, which have their own flushes.
Hence the CB cache is always up to date.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_meta_fast_clear.c | 2 --
1 file chan
I think we should only flush right before an action (draw/dispatch etc.),
as otherwise it is too easy to issue redundant flushes.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 4
src/amd/vulkan/radv_meta_fast_clear.c | 1 -
src/amd/
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Mar 7, 2017 at 8:01 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> These really are only supported for vertex buffers.
>
> Signed-off-by: Dave Airlie <
Less IFETCH latency on misses. Shader code is write once read many,
so GTT doesn't make much sense anyway.
If it turns out to fragment the CPU visible VRAM too much, we can upload with
SDMA.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_pipeline.c
Thanks. Pushed.
- Bas
On Mon, Mar 13, 2017 at 2:28 PM, Alex Smith wrote:
> This must be set to ICD_LOADER_MAGIC by vkAllocateCommandBuffers, which
> was being done when allocating a new buffer but not when reusing an
> existing one in the cache. This would hit an
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 13, 2017 at 9:52 PM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> Ported from radeonsi, pointed out by Tom.
>
> "This prevents LLVM from using sext i
Reviewed-by: Bas Nieuwenhuizen <ba...@google.com>
On Wed, Mar 8, 2017 at 11:27 PM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> Jason has patches to add validation to this area, this should fix
> radv shaders.
>
&
Nice catch.
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 6, 2017 at 3:54 PM, Alex Smith <asm...@feralinteractive.com> wrote:
> If we have any pending flushes on the primary command buffer, these
> must be performed before executing the secondary buff
It has issues and the fix I'm working on is too complicated for stable,
so disable for now.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
CC: 13.0 17.0 <mesa-sta...@lists.freedesktop.org>
---
src/amd/vulkan/radv_image.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/sr
Still not sure we can support miptrees when sampling from
HTILE enabled textures.
Added the tcCompatible winsys stuff while I'm at it.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_radeon_winsys.h| 5 +
src/amd/vulkan/winsys/
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 16 +++-
src/amd/vulkan/radv_image.c | 3 +--
2 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Mar 7, 2017 at 1:30 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This fixes:
> dEQP-VK.pipeline.render_to_image.3d.huge.depth.r8g8b8a8_unorm
>
&
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Mar 7, 2017 at 12:33 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> Don't fast clear inside the meta loop as things get
> conf
v2: fix levelCount assert.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 16 +++-
src/amd/vulkan/radv_image.c | 3 +--
2 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 10 ++---
src/amd/vulkan/radv_device.c | 4 +-
src/amd/vulkan/radv_image.c | 83 ---
src/amd/vulkan/radv_meta_clear.c | 2 +-
src/amd/
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Tue, Mar 7, 2017 at 6:09 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> No idea what this does, but disabling it fixes a bunch
> of failing CTS test
Pushed.
On Sun, Mar 5, 2017 at 10:04 PM, Grazvydas Ignotas wrote:
> Mainly to avoid gcc's complains about uninitialized ptr and offset use
> later in that code.
>
> Signed-off-by: Grazvydas Ignotas
> ---
> src/amd/vulkan/radv_cmd_buffer.c | 7 ---
> 1
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/Makefile.sources| 1 +
src/amd/vulkan/radv_debug.h| 40 ++
src/amd/vulkan/radv_device.c | 4 ++-
src/amd/vulkan/radv_pri
So that we don't keep allocating BOs for the IBs and upload buffers.
We run some risk of memory increase with e.g. a bimodal size
distribution of command buffers, but I haven't noticed a significant
increase with dota2 and talos.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
s
On Sun, Mar 5, 2017 at 11:39 PM, Grazvydas Ignotas <nota...@gmail.com> wrote:
> On Mon, Mar 6, 2017 at 12:05 AM, Bas Nieuwenhuizen
> <b...@basnieuwenhuizen.nl> wrote:
>> On Sun, Mar 5, 2017 at 10:04 PM, Grazvydas Ignotas <nota...@gmail.com> wrote:
>>>
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 6, 2017 at 12:27 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> This isn't exposed in -pro, the hw docs say it is deprecated,
> so let's not bother wi
16 * layout->dynamic_offset_count,
> + 256, , ))
> + return;
We need to set cmd_buffer->record_fail on failure too. With that
fixed, this series is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
>
>
Don't flush multiple times if we clear multiple attachments. Also allows
doing the depth clear in parallel with the fast color clears.
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_meta_clear.c | 49 +---
1 file chang
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_private.h | 18 ++-
src/amd/vulkan/si_cmd_buffer.c | 70 ++
2 files changed, 54 insertions(+), 34 deletions(-)
diff --git a/src/amd/vulkan/radv_private.h b/src/amd/
Signed-off-by: Bas Nieuwenhuizen <ba...@google.com>
---
src/amd/vulkan/radv_cmd_buffer.c | 8
src/amd/vulkan/radv_meta_clear.c | 4 ++--
2 files changed, 6 insertions(+), 6 deletions(-)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index 16c3f
Series is
Reviewed-by: Bas Nieuwenhuizen <b...@basnieuwenhuizen.nl>
On Mon, Mar 6, 2017 at 12:29 AM, Dave Airlie <airl...@gmail.com> wrote:
> From: Dave Airlie <airl...@redhat.com>
>
> Signed-off-by: Dave Airlie <airl...@redhat.com>
> ---
> src/amd/common
601 - 700 of 2350 matches
Mail list logo