fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-21 22:20:14 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1
185233 commits 64 branches 914 tags 1 GiB
Commit graph

3110 commits

Author SHA1 Message Date
Caio Oliveira
e8dd8b461e intel/brw: Pull fixup_nomask_control_flow out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
1bdfa5230f intel/brw: Pull lower_logical_sends out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
422b4bead6 intel/brw: Pull lower_barycentrics out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
af2b3c478c intel/brw: Pull lower_simd_width out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
14954cbd58 intel/brw: Pull lower_pack out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
f6b82b0270 intel/brw: Pull lower_constant_loads out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
a1c35563ae intel/brw: Pull register_coalesce out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
ce10130f09 intel/brw: Pull remove_extra_rounding_modes out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
73fe658456 intel/brw: Pull eliminate_find_live_channel out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
4f314f89f7 intel/brw: Pull opt_zero_samples out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
838d6d5cd2 intel/brw: Pull opt_split_sends out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
8dcbdc8fac intel/brw: Pull split/compact virtual_grf opts out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
1947a38680 intel/brw: Pull opt_algebraic out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
d849c2ecff intel/brw: Pull redundant_halt out of fs_visitor 
And call it "remove redundant halts".

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
96c4aa8545 intel/brw: Pull peephole_sel out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
10489b418c intel/brw: Pull bank_conflicts out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
13c312431c intel/brw: Pull opt_cse out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:25 +00:00
Caio Oliveira
4f09ad9dee intel/brw: Pull opt_combine_constants out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:24 +00:00
Caio Oliveira
59bff8adf4 intel/brw: Pull dead_code_eliminate out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:24 +00:00
Caio Oliveira
1bd175f458 intel/brw: Pull opt_saturate_propagation out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:24 +00:00
Caio Oliveira
dc33a8fb06 intel/brw: Pull opt_cmod_propagation out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:24 +00:00
Caio Oliveira
6a3329a6c4 intel/brw: Pull opt_copy_propagation out of fs_visitor 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:24 +00:00
Caio Oliveira
0b73d163d4 intel/brw: Remove Gfx8- passes from optimize() 
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/26887>
 2024-02-26 20:54:24 +00:00
Lionel Landwerlin
abeac8cf96 intel/nir: only consider ray query variables in lowering 
Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: mesa-stable
Reviewed-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27752>
 2024-02-24 12:56:30 +00:00
Caio Oliveira
373130a66c intel/compiler: Remove has_render_target_reads from wm_prog_data 
This was used only by the classic i965 driver.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27772>
 2024-02-24 02:34:59 +00:00
Caio Oliveira
10230d2eec intel/brw: Assert Gfx9+ 
Acked-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27768>
 2024-02-24 02:10:56 +00:00
Caio Oliveira
40437bea86 intel/brw: Remove pass test cases for Gfx8- 
And update the mock devinfo versions to be at least Gfx9.

Acked-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27768>
 2024-02-24 02:10:56 +00:00
Caio Oliveira
8c3165f0ff intel/brw: Remove EU validation tests for Gfx8- 
Acked-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27768>
 2024-02-24 02:10:56 +00:00
Caio Oliveira
a5b67d4bea intel/brw: Remove EU compaction tests for Gfx8- 
Acked-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27768>
 2024-02-24 02:10:56 +00:00
Caio Oliveira
9bc670509c intel/brw: Remove assembler tests for Gfx8- 
Acked-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27768>
 2024-02-24 02:10:56 +00:00
Caio Oliveira
80de55801e intel/elk: Remove tests for Gfx9+ 
These are covered in the regular "brw" compiler.

Acked-by: Ivan Briano <ivan.briano@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27770>
 2024-02-24 01:41:31 +00:00
Caio Oliveira
32db7a9533 intel/elk: Use anonymous namespace in fs_combine_constants 
Certain GitLab CI build use a combination of LTO and -Werror=odr that
will fail if both ELK and BRW share the same names for those helpers,
so wrap the ELK ones around anonymous namespace.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:32 +00:00
Caio Oliveira
2bc18fe46f intel-clc: Use correct set of nir_options when building for Gfx8 
Use the correct set of nir_options when building for Gfx8.  Note this is
only used in the NIR codepath.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:32 +00:00
Caio Oliveira
c83f92087b intel/elk: Move nir_options to its own c/h file pair 
This will allow intel-clc tool to use the ELK nir_options in its NIR
codepath without having to link with the entire ELK compiler.  That way
an Anv only build doesn't need to compile ELK.

Iris uses that codepath for Gfx8.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:32 +00:00
Caio Oliveira
8abfbc9cca intel/meson: Rename libintel_compiler to libintel_compiler_brw 
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:32 +00:00
Caio Oliveira
4866649052 intel/elk: Create separate header for opcodes 
This allow us to not include elk_eu_defines.h (mostly internal
information) in elk_isa_info.h (that's used by other modules).
This will help when using both brw and elk in Iris later.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
77f52417c5 intel/elk: Don't include elk_eu_defines.h in elk_nir.h 
Those definitions are not needed by the drivers (client code), so
reduce the exposure.  This will help when using both brw and elk in
Iris later.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
6648e0ebd3 intel/elk: Rename symbols 
Either replace the BRW prefix with ELK or add an extra ELK prefix.  Used
the following sed script to perform the renames in this patch:

```
    # Simple prefix changes.
    s/\<BRW_/ELK_/g
    s/\<brw_/elk_/g
    s/nir_to_brw/nir_to_elk/g
    s/\<as_brw_reg\>/as_elk_reg/g
    s/\<_brw_/_elk_/g

    # Add prefix to various symbols.
    #
    # Initially I've considered using C++ namespaces here, but in various
    # cases the structs or functions had to be also visible from C code.
    # So added explicit prefix instead.
    s/\<backend_instruction/elk_\0/g
    s/\<backend_reg/elk_\0/g
    s/\<backend_shader/elk_\0/g
    s/\<bblock_t\>/elk_\0/g
    s/\<bblock_link\>/elk_\0/g
    s/\<cfg_t\>/elk_\0/g
    s/\<fs_visitor\>/elk_\0/g
    s/\<fs_reg\>/elk_\0/g
    s/\<fs_instruction_scheduler\>/elk_\0/g
    s/\<vec4_instruction_scheduler\>/elk_\0/g
    s/\<instruction_scheduler\>/elk_\0/g
    s/\<schedule_node\>/elk_\0/g
    s/\<schedule_node_child\>/elk_\0/g
    s/\<\([a-z]*_\)\?thread_payload\>/elk_\1thread_payload/g
    s/\<fs_generator\>/elk_\0/g
    s/\<fs_inst\>/elk_\0/g
    s/\<fs_reg_alloc\>/elk_\0/g
    s/\<disasm_info\>/elk_\0/g
    s/\<gfx._math\>/elk_\0/g
    s/\<gfx7_block_read_scratch\>/elk_\0/g
    s/\<gfx6_IF\>/elk_\0/g
    s/\<gfx9_fb_READ\>/elk_\0/g
    s/\<gfx6_resolve_implied_move\>/elk_\0/g

    # Opcodes.
    s/\<opcode op\>/elk_\0/g
    s/\<opcode mov_op\>/elk_\0/g
    s/\<opcode opcode\>/elk_\0/g
    s/enum opcode\>/enum elk_opcode/g
    s/static opcode\>/static elk_opcode/g
    s/\<opcode elk_op/elk_opcode elk_op/g
    s/struct opcode_desc/struct elk_opcode_desc/g
    s/NUM_BRW_OPCODES/NUM_ELK_OPCODES/g
    s/\<.._OPCODE_/ELK_\0/g
    s/\<T.._OPCODE_/ELK_\0/g
    s/\<VEC4_OPCODE_/ELK_\0/g
    s/\<VEC4_...\?_OPCODE_/ELK_\0/g
    s/\<SHADER_OPCODE_/ELK_\0/g

    # Remaining specific cases.
    s/\<wm_prog_data_barycentric_modes\>/elk_\0/g
    s/\<encode_slm_size\>/elk_\0/g
    s/\<intel_calculate_slm_size\>/elk_\0/g
    s/\<gfx6_gather_sampler_wa\>/elk_\0/g
    s/\<is_3src\>/elk_\0/g
    s/\<WA_/ELK_\0/g
    s/\<conditional_modifier\>/elk_\0/g
    s/\<pred_ctrl_align16\>/elk_\0/g
    s/\<shuffle_from_32bit_read\>/elk_\0/g
    s/\<shuffle_src_to_dst\>/elk_\0/g
    s/\<setup_imm_..\?\>/elk_\0/g

    s/\<opt_predicated_break\>/elk_\0/g
    s/\<has_bank_conflict\>/elk_\0/g
    s/\<dead_control_flow_eliminate\>/elk_\0/g

    s/\<disasm_new_inst_group\>/elk_\0/g
    s/\<disasm_initialize\>/elk_\0/g
    s/\<dump_assembly\>/elk_\0/g
    s/\<disasm_insert_error\>/elk_\0/g
    s/\<disasm_annotate\>/elk_\0/g

    s/\<enum lsc_opcode\>/enum elk_lsc_opcode/g
    s/\<lsc_opcode_/elk_lsc_opcode_/g
    s/\<lsc_aop_[a-z_]\+\>/elk_\0/g

    s/\<type_size_vec4\>/elk_\0/g
    s/\<type_size_dvec4\>/elk_\0/g
    s/\<type_size_xvec4\>/elk_\0/g
    s/\<type_size_[a-z4]\+_bytes\>/elk_\0/g

    s/\<gfx12_systolic_depth\>/elk_\0/g
```

Acked-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
f677485ad4 intel/elk: Rename C++ namespace 
Acked-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
ed89a42f7a intel/elk: Update doxygen-like file comments 
Not sure if we use them, but updated.

Acked-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
783f3d0b4f intel/elk: Rename header guards 
Acked-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
e6022281f2 intel/elk: Rename files to use elk prefix 
Acked-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
b02712efc2 intel/elk: Remove DPAS lowering 
This is meant for Gfx9+.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
b743ab7acc intel/elk: Remove stages not used in Gfx8- 
OpenCL, Mesh and RayTracing stages are not supported, so removing them
and related code.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:31 +00:00
Caio Oliveira
8a57012ff4 intel/elk: Use common code in intel/compiler 
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:30 +00:00
Caio Oliveira
dcf29202d4 intel/elk: Remove a bunch of files that don't apply for Gfx8- 
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:30 +00:00
Caio Oliveira
06b553f02c intel/elk: Remove compiler specific devinfo hash 
This more coarse-grained hash information for compiler (vs. full
devinfo), used only by Iris and Anv, and relevant for more recent
platforms.  Remove it from elk.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:30 +00:00
Caio Oliveira
0083585fc5 intel/elk: Compile ELK library, tests and tools 
For now is not linked to any driver.  The tools were renamed to use elk
prefix to avoid conflicting with the brw ones.  The run-test.py script
was also updated due to that change.

Before the new compiler can be linked together with the old (going to be
done for Iris and other tools), the symbol conflicts need to be fixed
first.  This will happen in a later commit.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:30 +00:00
Caio Oliveira
d44462c08d intel/elk: Fork Gfx8- compiler by copying existing code 
Based on code from commit c3ceec6cd8.

Acked-by: Ian Romanick <ian.d.romanick@intel.com>
Acked-by: Kenneth Graunke <kenneth@whitecape.org>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27563>
 2024-02-24 00:24:30 +00:00
Kenneth Graunke
c12300844d intel/fs: Don't rely on CSE for VARYING_PULL_CONSTANT_LOAD 
In the past, we didn't have a good solution for combining scalar loads
with a variable index plus a constant offset.  To handle that, we took
our load offset and rounded it down to the nearest vec4, loaded an
entire vec4, and trusted in the backend CSE pass to detect loads from
the same address and remove redundant ones.

These days, nir_opt_load_store_vectorize() does a good job of taking
those scalar loads and combining them into vector loads for us, so we
no longer need to do this trick.  In fact, it can be better not to:
our offset need only be 4 byte (scalar) aligned, but we were making it
16 byte (vec4) aligned.  So if you wanted to load an unaligned vec2,
we might actually load two vec4's (___X | Y___) instead of doing a
single load at the starting offset.

This should also reduce the work the backend CSE pass has to do,
since we just emit a single VARYING_PULL_CONSTANT_LOAD instead of 4.

shader-db results on Alchemist:
- No changes in SEND count or spills/fills
- Instructions: helped 95, hurt 100, +/- 1-3 instructions
- Cycles: helped 3411 hurt 1868, -0.01% (-0.28% in affected)
- SIMD32: gained 5, lost 3

fossil-db results on Alchemist:
- Instrs: 161381427 -> 161384130 (+0.00%); split: -0.00%, +0.00%
- Cycles: 14258305873 -> 14145884365 (-0.79%); split: -0.95%, +0.16%
- SIMD32: Gained 42, lost 26

- Totals from 56285 (8.63% of 652236) affected shaders:
- Instrs: 13318308 -> 13321011 (+0.02%); split: -0.01%, +0.03%
- Cycles: 7464985282 -> 7352563774 (-1.51%); split: -1.82%, +0.31%

From this we can see that we aren't doing more loads than before
and the change is pretty inconsequential, but it requires less
optimizing to produce similar results.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27568>
 2024-02-20 23:16:27 -08:00