AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
187,766 Commits 1 Branch 0 Tags
24b6a047eee5b57863a4dcceef72bb344794d8c5
Commit Graph

9234 Commits

Author SHA1 Message Date
Georg Lehmann b44f97a7ba nir: don't try to optimize exclusive min/max scan to inclusive 
SPIR-V rules for fmax/fmin scans are *very* stupid.
The required identity is Inf instead of NaN but if one input
is NaN, the other value has to be returned.
This means for invocation 0:
min(subgroupExclusiveMin(NaN), NaN) -> Inf
subgroupInclusiveMin(NaN) -> undefined (NaN for any sane backend)

Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27536>
 2024-04-11 00:03:13 +00:00
Konstantin Seurer 85e840786c nir: Add lavapipe ray tracing intrinsics 
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28187>
 2024-04-09 07:13:01 +00:00
Konstantin Seurer edc8e011eb nir/serialize: Encode data for temporaries 
the location has to be preserved when lowering them to scratch using
nir_lower_vars_to_explicit_types and nir_lower_explicit_io.

cc: mesa-stable

Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28187>
 2024-04-09 07:13:01 +00:00
Konstantin Seurer 1ff8659f41 nir/print: Fix printing booleans with bit_size>1 
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28187>
 2024-04-09 07:13:01 +00:00
Paul Gofman bd189dbd77 glsl: allow out arrays in #110 with allow_glsl_120_subset_in_110 
CC: mesa-stable
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28439>
 2024-04-09 01:29:23 +00:00
Connor Abbott 42ddbf4ebe nir/divergence_analysis: Fix load_view_index divergence in VS 
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28573>
 2024-04-08 23:20:08 +00:00
Connor Abbott 223e3727d5 nir/divergence_analysis: Add uniform_load_tears option 
This "tear" is similar to the original concept of loads/stores tearing,
but across invocations in a wave instead of bytes. Qualcomm seems to
have this problem, at least for some GPUs. This fixes
spec@arb_shader_storage_buffer_object@execution@ssbo-atomiccompswap-int
on a630 once we start relying on divergence analysis for computing
reconvergence properties. For backends that have readFirstInvocation(),
it should be possible to fix the problem by inserting
readFirstInvocation() instead, but a5xx doesn't have it so we can't rely
on it in freedreno.

Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28573>
 2024-04-08 23:20:08 +00:00
Connor Abbott 291e6d2940 nir/divergence_analysis: Add ir3-specific intrinsics 
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28573>
 2024-04-08 23:20:08 +00:00
Pavel Ondračka 96479794a5 nir/lower_vec_to_regs: always set cursor before inserting decl_reg 
Otherwise we can end with invalid curson in some cases, specifically,
if we encounter an identity vector and thus neither register
declaration nor swizzle is inserted, it can lead to invalid cursor and a
crash later.

Suggested by: Alyssa Rosenzweig <alyssa@rosenzweig.io>

Signed-off-by: Pavel Ondračka <pavel.ondracka@gmail.com>
Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10905
Reviewed-by: Christian Gmeiner <cgmeiner@igalia.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28581>
 2024-04-08 21:03:46 +00:00
Rhys Perry 543ca160a5 nir,aco: add test intrinsics 
These don't really do anything. They're just a source and user of SSA
defs.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28301>
 2024-04-08 18:38:39 +00:00
Rhys Perry 863718d62e nir: add nir_remove_after_cf_node helper 
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Daniel Schürmann <daniel@schuermann.dev>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28301>
 2024-04-08 18:38:39 +00:00
Timothy Arceri f60956d002 glsl: move check_explicit_uniform_locations() to NIR linker 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri 2a5ae739ea glsl: move link_assign_subroutine_types() to the nir linker 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri baf13bff29 nir: add subroutine fields to nir_function 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri 5ea15ded2f glsl: switch verify_subroutine_associated_funcs() to nir 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri 1a78e9a7e8 glsl: move some linking calls to gl_nir_link_glsl() 
This is a more logical place for them and will avoid us adding more
function validtion calls to the st glsl to nir file in future
patches.

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri dbaa90200c glsl: remove now unused glsl ir block validation 
We now use a NIR based implementation instead.

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri 07078d4b8e glsl: remove now unused resize_tes_inputs() 
This has be replaced by a NIR based implementation of this function.

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri c2066d4685 glsl: call new nir resize_tes_inputs() pass 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri d1c11d957a glsl: switch to NIR block validation 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri aea1265470 glsl: add nir implemenation of block validation 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri 20978004a7 glsl: don't remove redefined per vertex block 
Commit 719bf30165 added this removal code with the following
justification:

"The GLSL 4.10 rules for redeclaration of built-in interface blocks
 (which we've chosen to regard as clarifications of GLSL 1.50) only
 require gl_PerVertex blocks to match in shaders that actually use
 those blocks.  The easiest way to implement this is to detect
 situations where a compiled shader doesn't refer to any elements of
 gl_PerVertex, and remove all the associated ir_variables from the
 shader at the end of ast-to-ir conversion."

However the intention is to avoid matching a redeclared block with
gl's default block if unused. We are still required to do block
matching in the shader should the block be redeclared, even if unused.
So with this change we only remove the block if it is both unused
and not redeclared.

The existing glsl IR code managed to avoid failing CTS tests for this
due to seemingly magical or hacky use of the symbol table but fixing
it will make things much clearer, and also allow a nir version of
this validation in a following patch.

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:04 +00:00
Timothy Arceri c46827d52a nir: add variable field from_ssbo_unsized_array 
This will be used for validation in following patches

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:03 +00:00
Timothy Arceri ab801a1a27 glsl: add resize_tes_inputs() to the nir linker 
We will switch from the GLSL IR version to this in a following patch.

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:03 +00:00
Timothy Arceri a18e06f694 nir: add implicit_sized_array data field 
Will be used in following patches for glsl linking validation.

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:03 +00:00
Timothy Arceri f24e6a5062 nir: add max_array_access data field 
Will be used in following patches for glsl linking validation.

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28538>
 2024-04-06 05:56:03 +00:00
Jesse Natalie acbf3ad1fb glsl: Use a stable attr sort for VS in / FS out 
This is a perpetual bug that hits Windows. In the MSVC CRT, qsort
is unstable, where the glibc qsort is stable. So apps run fine on
Windows IHV drivers, and on Linux Mesa drivers, and only break down
when running on Windows Mesa drivers.

Closes: https://gitlab.freedesktop.org/mesa/mesa/-/issues/10922
Reviewed-by: Sil Vilerino <sivileri@microsoft.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: mesa-stable
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28586>
 2024-04-05 20:35:05 +00:00
Mike Blumenkrantz a9d366085c nir/gather_info: fix gathering for compact arrayed builtins 
these are treated as vecs, but depending on driver caps num_slots
may or may not equal the array size

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28554>
 2024-04-05 17:56:28 +00:00
Mike Blumenkrantz 843614f45a nir: add compact_arrays to nir_shader_compiler_options 
this is used to inform passes about semantics for compact array builtins

Reviewed-by: Jesse Natalie <jenatali@microsoft.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28554>
 2024-04-05 17:56:28 +00:00
Rhys Perry 08903bbe89 nir: add mqsad_4x8, shfr and nir_opt_mqsad 
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26251>
 2024-04-05 11:01:39 +00:00
Mike Blumenkrantz 56cda23459 nir/dominance: fix comment 
this function doesn't exist

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:43 +00:00
Mike Blumenkrantz 0851c30d16 nir/texcoord_replace: fix scalarized io handling 
if a texcoord load only loads some components, only those components
should be replaced

cc: mesa-stable

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:43 +00:00
Mike Blumenkrantz f3d9a2e607 nir/lower_point_size_mov: fix for lowered io 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz 0faeeb6347 nir/lower_point_size_mov: rework. 
this was weirdly complex

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz 30415fe835 nir/lower_clip_disable: fix for lowered io 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz eb190a4e8b nir/lower_drawpixels: fix for lowered io 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz c48f580a49 nir/lower_two_sided_color: rework for lowered io 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz 9af9446127 nir/lower_alpha_test: fix use with lowered io 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz 842ac5059a nir/lower_flatshade: fix with lowered io 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz 712bf98dd8 nir/lower_flatshade: break out location checking 
no functional changes

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz 56d7dad19c nir/lower_clamp_color_outputs: fix use with lowered io 
if io is lowered, variables cannot be used

cc: mesa-stable

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Mike Blumenkrantz 056fe1dc6a glsl: set PSIZ bit in outputs_written when injecting a 1.0 psiz write 
gather_info may not be called before something checks this value

cc: mesa-stable

Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28463>
 2024-04-04 18:58:42 +00:00
Marek Olšák 772149b15a nir/opt_varyings: handle load_input_vertex 
Explicit interpolation just loads raw vertex data as-is and lets the FS do
the interpolation manually.

This adds handling of nir_intrinsic_load_input_vertex, which has 2 different
behaviors: undefined vertex ordering and strict vertex ordering.

- dead IO removed correctly
- constants and uniform expressions are propagated normally
- outputs are deduplicated within their own category (strict and non-strict)
- outputs used by explicit interpolation are never treated as "convergent"
- backward inter-shader code motion is skipped
- compaction has 2 new types of vec4 slots:
    - mixed 32-bit and 16-bit explicit strict (sharing the same vec4)
    - mixed 32-bit and 16-bit explicit non-strict (sharing the same vec4)

Reviewed-by: Timur Kristóf <timur.kristof@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28247>
 2024-04-04 01:25:06 +00:00
Ganesh Belgur Ramachandra 5b301e74ed compiler,glsl: fix warning when -finstrument-functions is used 
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28411>
 2024-04-04 00:03:53 +00:00
Mike Blumenkrantz ba52fb0eca glsl: handle xfb resources for spirv before running varying opts 
this otherwise breaks lowered io

cc: mesa-stable

Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28529>
 2024-04-03 19:28:18 +00:00
Mike Blumenkrantz 3d43f8c1a1 nir/lower_wpos_ytransform: fix for lowered io 
this should now handle scalarized+lowered io correctly

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28462>
 2024-04-03 16:42:46 +00:00
Mike Blumenkrantz 1c527dab96 nir/lower_wpos_ytransform: scalarize emit_wpos_adjustment 
should be no functional changes

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28462>
 2024-04-03 16:42:46 +00:00
Mike Blumenkrantz a9e023ed94 nir/lower_wpos_ytransform: update comment to reflect variable usage 
Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28462>
 2024-04-03 16:42:46 +00:00
Mike Blumenkrantz e871424b5e nir/lower_wpos_ytransform: reuse input zw components for fragcoord rewrite 
no functional changes, these are just unchanged

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28462>
 2024-04-03 16:42:46 +00:00
Mike Blumenkrantz e848d9b9cc nir/lower_wpos_ytransform: move new value load to start of function, reuse 
should be no functional changes

Acked-by: Marek Olšák <marek.olsak@amd.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28462>
 2024-04-03 16:42:46 +00:00