We need the traversal stack to saved/restored along with mem hits.
Total spill/fill is 256bytes.
We can potentially optimize this but we have to be very careful about
what state the query is in.
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: c78be5da30 ("intel/fs: lower ray query intrinsics")
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17396>
This function copies the potential hit from its memory location to the
committed hit location. A couple of fields got their bit offset wrong.
Fixes some CTS tests in dEQP-VK.ray_query.*
v2: Copy primitive/instance leaf pointers
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Fixes: 0465714790 ("intel/nir/rt: add more helpers for ray queries")
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17396>
Instead of trying to guess the interface style by solely checking
the OS name, check if a fake test program can be built and linked.
This will give more accurate result when FreeBSD and other systems
moved to the interface based on GNU qsort_r().
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Eric Engestrom <eric@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18527>
By default sb is disabled, to enable it one can
set
R600_DEBUG=nirsb
TGSI can be enabled by using
R600_DEBUG=use_tgsi
In this case "sb" will be enabled, but one can use
the TGSI backend without sb by setting
R600_DEBUG=use_tgsi,sb
Closes: #6858
Signed-off-by: Gert Wollny <gert.wollny@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18753>
the ->clear path checks render condition which isn't necessary.
Fixes dEQP-VK.conditional_rendering.draw.condition_host_memory_expect_noop_rp_clear.draw*
Reviewed-by: Brian Paul <brianp@vmware.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18686>
Pass the VA directly and rename to radv_make_texel_buffer_descriptor().
This helper will be re-used for future work.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18750>
It's not really useful.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18750>
For future work, and remove useless radv_device parameter.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18750>
This feature is only concerned with buffers bound through a descriptor
set. We are still keeping the code for this (disabled by default) since
it may be useful for debugging some scenarios.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18744>
Our implemention was bogus, it was only putting a cap on the offset
based on the aligned buffer size and this doesn't ensure the access
to the buffer happens within its valid range.
I think the only reason we have been passing the tests is that we
align all buffers sizes to 256B and the tests create buffers with a
size that is smaller than that (like 64B). When get the size of the
buffer from the shader, we get the actual bound range (so 64B in this
case) and by capping to that we don't ensure the access will stay
within that range, but we ensure it will stay within the underlying
memory bound to the buffer (256B), and this is fine by the spec,
however, I think if the actual buffer range was the same as the
underlying allocation we would fail the tests.
A valid behavior for robust buffer access on an out-of-bounds access
is to return any valid bytes within the buffer, so we can just
make that offset 0.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18744>
When display list loopback path is hit the major performance
drop during glCallList() happens due to constantly mapping and
unmapping the buffer.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17285>
When deciding on workarounds that apply to GLSLang we should also check
for ShaderC (when using glslang). Add a helper to do that and reuse in
the three places we check for it.
This patch will now consider shaderc for wa_glslang_cs_barrier workaround.
Suggested by Timur Kristóf.
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18442>
No need to call it at the end of every block, since we already
know and identified the blocks terminating with returns.
Suggested by Jason Ekstrand.
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18442>
glslang had a bug and didn't actually emit it as a terminating
instruction, therefore this went unnoticed thus far. See
https://github.com/KhronosGroup/glslang/issues/3020.
Modified by Caio Oliveira to emit the corresponding NIR intrinsic
right before the halt.
Fixes: 7d1bcf1f55 ("spirv, nir: Handle EmitMeshTasksEXT opcode.")
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18442>
This will be used later to handle branches instructions that have
parameters in them, e.g. SpvOpEmitMeshTasksEXT. No behavior change in
this commit.
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18442>
The check in d3d12_has_video_encode_support was returning no encode support unless
the driver supported D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_UNIFORM_PARTITIONING_SUBREGIONS_PER_FRAME
when it should also return support when the driver reports only full frame support
(ie. D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_FULL_FRAME)
Reviewed-by: Giancarlo Devich <gdevich@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18763>
instead of incurring all the overhead of tracking lifetimes for these,
it makes more sense to just let them delete whenever and then store
the vk object onto its parent image/buffer to be deleted when that gets
freed, as the parent object's lifetime will always be >= the view's lifetime
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18664>
this uses a revised version of radeonsi's buffer list array to track
resources for lifetime/usage management, massively improving performance
by avoiding hash table lookups
Reviewed-by: Adam Jackson <ajax@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18664>
