pipe_stream_output_info is built from nir_xfb_info, why not just use
nir_xfb_info directly.
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/20015>
XFB queries need to be enabled with NGG streamout and VS/TES.
Previously, the NGG lowering code relied on has_prim_query for XFB.
This fixes failures with RADV_PERFTEST=ngg_streamout on GFX10.3 with
the vkd3d-proton testsuite. Vulkan CTS is missing TES tests with XFB
queries apparently.
Cc: 22.3 mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/19493>
Use lds base load intrinsics in nir ngg lowering to get layout, left
its calulation to driver.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18832>
radeonsi need to use packed driver location for all outputs,
while radv need to use VARYING_SLOT_*. To meet both drivers'
needs.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/18832>
radeonsi may disable it. gfx_level will also be used by latter
vertex param export when gfx11.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17457>
Port from radeonsi.
Works on both GFX11 and GFX10. Although GFX10 can do atomic
GDS add on all threads, now we just disable the NGG streamout
for GFX10, so it's OK.
There's a difference for the GFX11 implementation with radeonsi
that we do all 4 buffer/stream info calc on a single thread.
It's just because this is simple, we need to update GDS on a
single thread anyway, and streamout is not that performance
critical to loss a small amount of instruction. We may change
to a better implementation when using register based streamout.
When streamout enabled, ES threads need to save all vertex
attributes to LDS besides position. This is because we don't
know where in the streamout buffer to export the attributes to
and wheter there are space in the streamout buffer.
Streamout is done in primitives, so we need to check if there
is space and where the current primitive should be written to
by GDS atomic add, then in GS threads do the streamout.
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17654>
Port from radeonsi.
Besides vertex position based primitive culling, clipdist
attribute can also be used to cull a primitive. Normally
it's used by fixed-pipeline, but when NGG we can treate it
as a culling condition to filter out invisible primitive
before fixed-pipeline.
There are two kinds of clipdist:
1. user define a clip plane explicitly by glClipPlane(),
fixed-pipeline calculate with vertex position to get
clipdist, then cull. This is the legacy way.
2. Now GLSL define gl_ClipDistance/gl_CullDiatance so that
user can calculate clipdist in any way he like.
This implementation support both way.
Acked-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17651>
Port from radeonsi.
Cull primitive after GS thread and before final vertex/primitive
export. GS culling is like VS/TES culling which read out saved
vertex positions of a primitive from LDS then call the primitive
culling algorithm to check whether it's visiable or not, only
passed primitives will be exported.
Unlike the VS/TES culling that read vertex index of a primitive
from VGPRs as shader args, GS will set a primitive complete flag
for each last vertex of a primitive in LDS, so that vertex thread
know the previous 1/2/3 vertex can form a primitive and do primitive
culling.
Acked-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17651>
Make accept_func optional, and return accpect result for caller
react when primitive is rejected.
This is for GS culling.
Acked-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17651>
For primitives which are rejected based on only W and face, this
will reduce the number of executed branches.
Fossil DB stats on Navi 21:
Totals from 60918 (45.16% of 134906) affected shaders:
CodeSize: 160330564 -> 160086644 (-0.15%)
Instrs: 30477385 -> 30477916 (+0.00%); split: -0.00%, +0.00%
Latency: 139802763 -> 139587915 (-0.15%); split: -0.15%, +0.00%
InvThroughput: 21198444 -> 21184261 (-0.07%); split: -0.07%, +0.00%
SClause: 749811 -> 749810 (-0.00%)
Copies: 2701482 -> 2762930 (+2.27%); split: -0.00%, +2.28%
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17870>
radeonsi may add extra dword to the stride, so let's pass it
directly.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16788>
This is used by radeonsi to save some lds space when all LS output
is passed by register.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>
Used by radeonsi and radv to reflect true wave size used, not minimal size.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>
radeonsi won't emit tess factor in the lower pass, need to keep
the output for llvm backend to pass it as parameter. This is used
by radeonsi for an optimization to save LDS write.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>
radeonsi load this from SGPR arg, can't use static value because TCS output
and TES input may not match (TCS output is not a key for TES) and
determined in runtime.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16705>
This will be used for radeonsi to map common I/O location to fixed
slots agreed by different shader stages.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16418>
For radeonsi which pass this value by argument.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16418>
Unfortunately, radv_lower_multiview is not suitable for mesh shaders
because it can't know the mapping between API mesh shader
invocations and output primitives.
Additionally, when lowering view id to layer, it must be created
as a per-primitive PS input.
Fixes: d32656bc65
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/16504>
Task shaders store their output payload to VRAM where mesh
shaders read from. There are two ring buffers:
1. Draw ring: this is where mesh dispatch sizes and
the ready bit are stored.
2. Payload ring: this is where the optional payload
is stored (up to 16K per task workgroup).
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/14929>
Lower mesh shader outputs to shared memory.
At the end of the shader, read the outputs from shared memory
and export their values as NGG expects.
We allocate separate shared memory (LDS) areas for per-vertex,
per-primitive outputs, primitive indices, primitive count.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13580>
The result is about +5-ish fps in Doom Eternal.
It turns out that the location of position exports matters more
than we thought, and it's actually better to keep them at the bottom
for culling shaders rather than schedule it up to the top.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10525>
Culling is traditionally done by the rasterizer, but that
can be a bottleneck when an app creates a large number
of primitives. Eg. a lot of tiny triangles reduce the
rasterziation efficiency.
NGG makes it possible for the shader to check primitives
and delete those that it can prove are not needed.
After this is done, we have to repack the surviving invocations
so they remain compact. This also saves bandwidth, because
some memory loads are only executed by those vertices that
survived the culling.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10525>
The algorithms were originally implemented by Marek Olšák,
hence the copyright to AMD.
This commit just ports the LLVM based implementation to NIR,
using the new intrinsics added earlier.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/10525>
