This reduces GLSL compile times with the gallium noop driver by 0.6%.
This might decrease register usage and do less code reordering because
nir_lower_io_vars_to_temporaries is no longer called for inputs, which
moved most input loads to the top.
radeonsi+ACO shader-db results are noise.
More uniforms are identified as inlinable.
TOTALS FROM ALL SHADERS (58138):
VGPRs: 2152680 -> 2158032 (0.25 %)
Code Size: 71008908 -> 71064812 (0.08 %) bytes
Max Waves: 916943 -> 916924 (-0.00 %)
Inline Uniforms: 6395 -> 6414 (0.30 %)
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36018>
This is a partial replacement for nir_lower_io_vars_to_temporaries.
It supports all input and output loads. It doesn't handle stores.
The motivation is to improve compile times.
The main differences compared to nir_lower_io_vars_to_temporaries are:
- it only lowers indirect loads to temps and doesn't touch direct loads
which improves compile times and removes the need for nir_lower_vars_to_ssa
afterward because indirect temp access can't be lowered to SSA
- it doesn't move all input loads to the top; it only moves those input
loads to the top whose indirect loads are lowered (which improves
register usage because direct loads are not moved)
- it doesn't have to deal with complexities of variables
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/36018>
It's done later in nir_lower_io_passes only for shader stages not
supporting indirect access.
Unfortunately we have add a hack into nir_lower_io_passes to get rid of
output loads. A later commit will remove it.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35945>
this lets the backend implement bounded loads (i.e. robust SSBOs) in a way
that's more clever than a full branch. similar idea to
load_global_constant_bound which should eventually be merged into this.
Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Job Noorman <job@noorman.info>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/35720>
Allocate space for the aliased region first, then allocate the
non-Aliased blocks in sequence after that.
SPV_KHR_workgroup_memory_explicit_layout extension added support for
having Blocks of workgroup (shared) memory, which include layout
decoration. For that extension all such blocks must be decorated with
Aliased.
SPV_KHR_untyped_pointers extension lifts that requirement, allowing
blocks that don't alias in workgroup memory. They are still explicitly
laid out.
The motivation is that untyped pointers provide a different
mechanism to obtain the same effect as the Aliased blocks. Instead of
having two Aliased variables with different types, have a single
variable and use an untyped pointer with a different type to access it.
This patch is a preparation for supporting untyped pointers.
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34139>
Move the calculation to nir_lower_vars_to_explicit_types(). This
consolidates the check of shader_info::shared_memory_explicit_layout
in a single place instead of in all drivers.
This is motivated by SPV_KHR_untyped_pointers. Before that extension
we had essentially two modes for shared memory variables
- No layout decorations in the SPIR-V, and both internal layout and
driver location was _given by the driver_.
- Explicitly laid out, i.e. they are blocks, and decorated with Aliased.
Because they all alias, we could assign them driver location directly
to the start of the shared memory.
With the untyped pointers extension, there's a third option, to be added
by a later commit
- Explicitly laid out, i.e. they are blocks, and NOT decorated
with Aliased. Driver location is _given by the driver_. Blocks
with and without Aliased can be mixed.
The driver location of multiple blocks that don't alias depend on
alignment that is driver-specific, which we can more easily do from
the nir_lower_vars_to_explicit_types() that already has access to
a function to obtain such value.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io> (hk)
Reviewed-by: Iago Toral Quiroga <itoral@igalia.com> (v3dv)
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> (anv/hasvk)
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com> (panvk)
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> (radv)
Reviewed-by: Rob Clark <robdclark@gmail.com> (tu)
Reviewed-by: Faith Ekstrand <faith.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34139>
load_attribute_pan is a panfrost-specific intrinsic for loading
vertex attributes. Takes explicit vertex and instance IDs which
we need in order to implement vertex attribute divisor with
non-zero base instance on v9+.
Passes which are used by panvk are modified to be aware of
load_attribute_pan.
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32039>
Signed-off-by: Benjamin Lee <benjamin.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31704>
In panvk we pass absolute view indices to the hardware, so we need to do
the conversion from compacted to absolute at some point. Emitting
absolute indices from nir_lower_multiview initially looks like the
simplest option, but nir_lower_io_to_temporaries will emit a write for
every element of array varyings. This results in unnecessary writes to
disabled views.
Signed-off-by: Benjamin Lee <benjamin.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31704>
This is needed for implementing multiview in panvk, where the address
calculation for multiview outputs is not well-represented by lowering to
nir_intrinsic_store_output with a single offset.
The case where a variable is both per-view and per-{vertex,primitive} is
now unsupported. This would come up with drivers implementing
NV_mesh_shader or using nir_lower_multiview on geometry, tessellation,
or mesh shaders. No drivers currently do either of these. There was some
code that attempted to handle the nested per-view case by unwrapping
per-view/arrayed types twice, but it's unclear to what extent this
actually worked.
ANV and Turnip both rely on per-view outputs being assigned a unique
driver location for each view, so I've added on option to configure that
behavior rather than removing it.
Signed-off-by: Benjamin Lee <benjamin.lee@collabora.com>
Reviewed-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31704>
This will need to be set to true when the GLSL linker lowers IO, which
can later be unlowered by st/mesa, and then drivers can lower it again
without load_interpolated_input. Therefore, it can't be a global
immutable option.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32229>
The main difference between load_global and load_global_constant is that
the latter can be reordered arbitrarily. If the access being lowered is
already tagged as being reorderable, then we can preserve that by using
the load_global_constant intrinsics instead of load_global. This gives
us more flexibility.
On Intel, this lets us use the load_global_constant_uniform_block_intel
intrinsic for doing convergent block loads in more cases. This nets us
significant reductions in spill/fills: Borderlands 3 on Lunarlake sees
spills/fills reduced by 53%. Alchemist sees a 13% reduction.
Improves performance of Borderlands 3 DX12 on Intel Battlemage by
around 44%. Improves Hogwarts Legacy by around 14%.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31995>
to improve CSE of load_barycentric_* and IO vectorization.
This is only for load_interpolated_input, which can never be FLAT.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31968>
When offset=0, the pass was a no-op but was setting the progress
flag which could cause infinite loops when this pass is going
to be added to gl_nir_opts.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31684>
The problem with radeonsi+ACO is that UBO loads from vec4 uniforms using
only 1 component always load all 4 components. This fixes that.
We are only interested in shrinking UBO and SSBO loads, but I added more
intrinsics because why not.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29384>
The ldc_nv and ldcx_nv intrinsics correspond to the index and bindless
forms of NVIDIA's LDC instruction, respectively. ldc_nv is pretty much
load_ubo without some of the unnecessary constant bits while ldcx_nv
takes a 64-bit bindless handle instead of an index. The other two give
us a little control over register allocation at the NIR level to ensure
that LDCX handles are placed in uniform registers.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29591>
For drivers that don't lower mediump shader inputs / outputs
to 16-bit, it's better to ignore the mediump flag completely,
letting mediump inputs / outputs work like normal 32-bit IO.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29435>
this creates a reference between variables and their access instrs
before the variables are deleted, which improves debugging
Acked-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28814>
Previously, this information would have been lost when the
shader has no I/O variables.
Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28489>
Will be needed to support push constants in
nir_lower_mem_access_bit_sizes().
Signed-off-by: Boris Brezillon <boris.brezillon@collabora.com>
Reviewed-by: Mary Guillemard <mary.guillemard@collabora.com>
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28175>
The varying linker isn't changed. The passes are executed after linking
varyings and before linking uniforms if nir->options->lower_io_variables
is true. nir_opt_varyings can move uniforms between shaders and cause them
to be DCE'd.
It requires moving IO deref lowering from st/mesa into the GLSL linker
and nir_opt_varyings should be added at the same time because IO deref
lowering alone would disable IO optimizations in st/mesa such as
compaction.
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26819>
this logic relies on constant indexing for compact arrays, but this is
frequently not the case for compact array builtins (e.g., gl_TessLevelOuter).
the usual strategy of lowering to temps isn't viable in TCS, which means
io lowering has to be able to handle indirect access to these builtins
without crashing
cc: mesa-stable
Reviewed-by: Dave Airlie <airlied@redhat.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27534>
