Lines Matching full:nir

34 - GL_NV_copy_depth_to_color for NIR
83 - Amber test NIR validation failed after spirv_to_nir
85 - Follow-up from "nir,spirv: Add generic pointers support"
90 - Follow-up from "st/mesa: Use nir-to-tgsi for builtins if the driver needs TGSI"
124 - nir: Mesa regression on Compute shader
168 - \[regression][bisected\] nir: nir_intrinsic_io_semantics assert failures in piglit
226 - nir/lower_io: don't reduce range if parent length is zero
265 - v3dv/pipeline: adding some nir-based linking
267 - v3dv: debug nir shader also after spirv_to_nir
275 - v3dv/pipeline: revamp nir lowering/optimizations passes
314 - nir: include texture query lod as one of the ops that requires a sampler
321 - v3dv/pipeline_cache: cache nir shaders
360 - nir/lower_io_to_scalar: update io semantics on per-component inst
396 - nir/lower_ssbo: Don't set align\_\* for atomics
489 - panfrost: Free cloned NIR shader
490 - panfrost: Free NIR of blit shaders
682 - pan/bi: Map NIR tex ops to Bifrost ops
752 - nir: Add SRC_TYPE to store_combined_output_pan
793 - nir: get rid of OOB dereferences in nir_lower_io_arrays_to_elements
823 - broadcom/compiler: use nir io semantics
895 - compiler/nir: Add new flags to lower pack/unpack split instructions
896 - nir: Fix i64tof32 lowering
905 - nir/glsl: Consider block interfaces as structs when it comes to size/align calculation
906 - nir: Expose the packed attribute attached to glsl_type objects
962 - panfrost: Don't leak NIR blend shaders
1051 - nir: Add nir_intrinsic_terminate and nir_intrinsic_terminate_if
1057 - nir: Use a switch in nir_lower_explicit_io_instr
1060 - nir: Add nir_intrinsic_{load,store}_deref_block_intel
1103 - nir: make tgsi_varying_semantic_to_slot(..) public
1107 - etnaviv: nir: do not run opt loop after nir_lower_bool_xxx(..)
1119 - nir/spirv: Add the option to keep ViewIndex as an input
1120 - nir/lower_input_attachments: Refactor to use an options struct
1121 - nir/lower_input_attachments: Support loading layer id as an input
1124 - nir/lower_input_attachments: Support loading layer id via gl_ViewIndex
1140 - nir: Add nir_lower_multiview pass
1142 - nir: Count i/o slots correctly for per-view variables
1143 - nir/lower_io_arrays: Fix xfb_offset bug
1144 - nir: Add per_view to IO semantics
1145 - nir: Handle per-view io in nir_io_add_const_offset_to_base()
1151 - intel/nir: Use nir control flow helpers
1152 - radv: Use nir control flow insertion helpers
1153 - ttn: Use nir control flow insertion helpers
1154 - nir/lower_returns: Use nir control flow insertion helpers
1155 - nir/opt_if: Remove open-coded nir_ssa_def_rewrite_uses()
1156 - nir/opt_if: Use early returns in opt_if_merge()
1158 - nir/lower_clip_cull: Store array size for FS inputs
1179 - nir,amd: remove trinary_minmax opcodes
1182 - ac/nir: implement nir_op_[un]pack_[64/32]_\*
1186 - nir/opt_algebraic: optimize unpack_half_2x16_split_x(ushr, a, 16)
1198 - ac/nir: implement nir_op_[un]pack_64_4x16
1199 - nir: add options to lower nir_op_pack_[64/32]_\* via nir_lower_alu_to_scalar()
1203 - nir/lcssa: consider loops with no back-edge invariant
1228 - anv/nir: Unify inputs_read/outputs_written between geometry stages
1238 - st/nir: Call st_glsl_to_nir_post_opts before interface unification
1239 - nir/lower_io: Eliminate oob writes and return zero for oob reads
1240 - nir/large_constants: Eliminate out-of-bounds writes to large constants
1241 - nir/lower_samplers: Clamp out-of-bounds access to array of samplers
1244 - nir/lower_returns: Append missing phis' sources after "break" insertion
1255 - gallivm/nir: add load push constant support
1263 - gallium/nir/tgsi: reindent some code in the nir->tgsi info (v2)
1264 - gallivm/nir: add imod support
1267 - gallium/nir/tgsi: add support for compact variables
1268 - gallivm/nir: fixup compact TCS variable storage.
1269 - gallivm/nir: split tex/sampler indirect offsets
1272 - llvmpipe/nir: lower some more packing at nir level.
1276 - gallium/nir/tgsi: fix nir->tgsi info conversion for samplers/image
1277 - gallivm/nir: lower tg4 offsets.
1278 - gallivm/nir: add indirect swizzle output loading support
1279 - gallivm/nir: add quantize to f16 support
1280 - gallivm/nir: fix const compact
1281 - gallivm/nir: lower frexp/ldexp
1282 - gallivm/nir: add subpass sampler type support
1287 - gallivm/nir: allow 64-bit arit ops
1288 - gallivm/nir: add some f16 support
1293 - vtn: add an option to create a nir library from spirv
1294 - clover/nir: add libclc lowering pass
1299 - gallivm/nir: fix up non 32-bit load stores
1300 - gallivm/nir: handle non-32-bit mul high
1303 - gallivm/nir: make sure to mask global reads.
1305 - gallivm/nir: fix non-32 bit find lsb/msb
1313 - clover/nir: add a constant folding pass before lowering mem const
1334 - gallivm/nir: handle dvec3/4 inputs properly.
1335 - gallivm/nir: fix vulkan vertex inputs
1458 - nir/opt_undef: Handle a couple more normal store intrinsics.
1459 - nir: Expand opt_undef to handle undef channels in a store intrinsic.
1460 - nir: Shrink store intrinsic num_components to the size used by the writemask.
1470 - nir: Make the nir_builder \*_imm helpers consistently handle bit size.
1471 - nir: Add nir_[iu]shr_imm and nir_udiv_imm helpers and use them.
1472 - nir: Add a lowering pass for backends wanting load_ubo with vec4 offsets.
1474 - nir/load_store_vectorizer: Clean up unit test swizzle assertions.
1479 - nir/opt_copy_prop_vars: Quiet valgrind warning about overlapping memcpy.
1480 - nir: Add a helper for general instruction-modifying passes.
1481 - nir/lower_vec_to_movs: Convert to use nir_shader_instructions_pass().
1482 - nir/opt_undef: Convert to use nir_shader_instructions_pass().
1483 - nir/lower_io_to_scalar: Convert to use nir_shader_instructions_pass().
1484 - nir/nir_lower_wrmasks: Use the nir_lower_instructions_pass() helper.
1485 - nir/lower_discard_to_demote: Use nir_shader_instructions_pass().
1499 - nir: Fix printing of individual instructions with io semantics.
1500 - nir: Look up the shader when printing a single instruction.
1505 - nir: Invalidate live SSA def information when making new SSA defs.
1506 - nir: Switch the indexing of block->live_in/out arrays.
1516 - nir: Add simplistic lowering for bany_equal/ball_inequal.
1517 - nir/opt_vectorize: Add a callback for filtering of vectorizing.
1526 - nir/load_store_vectorizer: Use more imm helpers in the tests.
1527 - nir/load_store_vectorizer: Add unit tests for alignment handling.
1528 - nir: Update the comment about nir_lower_uniforms_to_ubo()'s multiplier.
1529 - nir: Add a range_base+range to nir_intrinsic_load_ubo().
1530 - freedreno/ir3: Use the new NIR UBO ranges in UBO analysis.
1532 - nir: Use explicit deref information to provide real UBO ranges.
1559 - nir/lower_clip: Add i/o semantics for load/store intrinsics.
1562 - nir: Document a bit about how align_mul/offset work.
1563 - nir: Print the alignment information on casts.
1564 - nir/nir_lower_uniforms_to_ubo: Set better alignments on our new instructions.
1565 - nir/gl_nir_lower_buffers: Set up align_mul/offset on UBOs.
1566 - nir: Make the load_store_vectorizer provide align_mul + align_offset.
1567 - nir: Drop the high_offset argument to the load_store_vectorizer filter.
1568 - nir: Make nir_lower_ubo_vec4() handle non-vec4-aligned loads.
1613 - nir: Introduce nir_metadata_instr_index for nir_index_instr() being current.
1614 - nir: Replace nir_ssa_def->live_index with nir_instr->index.
1615 - nir: Add a block start/end ip to live instr index metadata.
1616 - nir: Add a call to get a struct describing SSA liveness per instruction.
1617 - nir: Add an option to not lower source mods for f64/u64/i64.
1618 - gallium: Add a nir-to-TGSI pass.
1620 - softpipe: Switch to using NIR as the shader format from mesa/st.
1633 - gallium/ntt: Add default compiler options for non-native-NIR drivers.
1634 - st/mesa: Drop the TGSI paths for PBOs and use nir-to-tgsi if needed.
1635 - st/mesa: Drop the TGSI paths for drawpixels and use nir-to-tgsi if needed.
1636 - nir: Only validate in passes that might have changed things.
1646 - mesa/st: Fix a use-after-free of the NIR shader stage.
1647 - st/nir: Fix the st->pbo.use_gs case.
1648 - st/nir: Drop setting interp mode on system values in builtins.
1781 - compiler/nir: make lowering global-id to local-id optional
1782 - nir: add iabs-lowering code
1784 - nir: fix const-cast warning on MSVC
1863 - nir: drop support for using load_alpha_ref_float
1864 - nir: drop unused alpha_ref_float
1911 - r600: Enable compute shaders for NIR code path
1912 - compiler/nir: rewrite lower_fragcoord_wtrans to use nir_lower_instructions
1913 - compiler/nir: extend lower_fragcoord_wtrans to support VARYING_SLOT_POS
1917 - r600/nir: fetch sources and split uniforms before emittting alu instructions
1923 - r600/sfn: Lower \*sign opcodes in nir
1929 - nir: Add option lower_uniforms_to_ubo
1937 - nir: remove ubo_r600 instrinsic since ubo_vec4 is used now
1960 - r600/sfn: replace hand-backed literal check by NIR function
1980 - compile/nir: Correct printing dest_type
2236 - v3dv: allow to create shader modules from NIR
2273 - v3dv: don't leak NIR code in pipelines
2391 - nir/glsl: add a glsl_ivec4_type() helper
2401 - nir/lower_io: add an option to lower interpolateAt functions
2402 - v3dv: lower interpolateAt functions in NIR and enable sample rate shading
2407 - nir: add a nir_get_ubo_size intrinsic
2494 - nir: Rename replicated-result dot-product instructions
2542 - mesa: add NV_copy_depth_to_color support for nir
2550 - nir: add shared/global atomics to nir_get_io_offset_src()
2558 - nir/algebraic: fold some nested comparisons with ball and bany
2581 - nir: Stabilize compact_components sort
2598 - iris: Add support for serialized NIR
2607 - nir: Add and use nir_foreach_block_unstructured helpers
2608 - nir/lower_goto_if: Document some data structures
2609 - nir/lower_goto_if: Clean up ralloc usage
2610 - nir/lower_goto_if: Use util/list instead of exec_list
2611 - nir/lower_goto_if: Rework handling of skip targets
2612 - nir/lower_goto_if: Rework some set union logic
2613 - nir/lower_goto_if: Sort blocks in select_fork
2614 - nir/lower_goto_if: Add a block_for_singular_set helper
2615 - nir/lower_goto_if: Replace a tripple loop with a double loop
2616 - nir/lower_goto_if: Add a route::outside set
2617 - nir/lower_goto_if: Add some debug prints
2619 - nir/builder: Make nir_get_ptr_bitsize take a nir_shader
2621 - clover/nir: Stop setting ubo_addr_format
2622 - clover/nir: Stop computing the global address format twice
2623 - clover/nir: Use the correct address mode for shared
2624 - nir: Initialize nir_ssa_def::live_index
2625 - nir/builder: Add a nir_iand_imm helper
2626 - nir/find_array_copies: Handle cast derefs
2627 - nir/large_constants: Handle incomplete derefs
2629 - nir/opt_large_constants: Fix a type/deref_type typo
2630 - nir: Add an LOD parameter to image_*_size
2637 - intel/nir: Allow splitting a single load into up to 32 loads
2640 - nir: Report progress properly in nir_lower_bool_to\_\*
2641 - intel/nir: Pass the nir_builder by reference in lower_alpha_to_coverage
2642 - intel/nir: Rewrite the guts of lower_alpha_to_coverage
2643 - intel/nir: Clean up lower_alpha_to_coverage a bit
2644 - nir: Use a switch in nir_inline_function_impl
2645 - nir: Take a variable remap parameter in nir_inline_function_impl
2647 - intel/nir: Lower things with \\> 4 components in lower_mem_access_bit_sizes
2650 - nir: Rename num_shared to shared_size
2651 - nir: Improve the comment on num_inputs and friends
2653 - nir: Add a new nir_var_mem_constant variable mode
2654 - nir: Add a load_global_constant intrinsic
2655 - nir/lower_io: Use the variable mode for load_scratch_base_ptr checks
2656 - nir/lower_io: Add a build_addr_for_var helper
2657 - nir/lower_io: Add support for nir_var_mem_constant
2658 - nir: Allow opt_large_constants to be run with constant_data_size \\> 0
2661 - nouveau/nir: Implement load_global_constant
2663 - clover/nir: Use nir_var_mem_constant for \__constant memory
2665 - nir/lower_explicit_io: Assert that compute address sizes match derefs
2667 - nir/clone: Add a helper for cloning most instruction types
2672 - intel/fs,vec4: Stuff the constant data from NIR in the end of the program
2678 - nir/builder: Add load/store_global helpers
2683 - intel/nir: Lower load_num_work_groups to 32-bit if needed
2686 - nir/lower_indirect_derefs: Add a threshold
2687 - intel/nir: Stop using nir_lower_vars_to_scratch
2688 - nir: Don't bail too early in lower_mem_constant_vars
2691 - nir/glsl: Add an explicit_alignment field to glsl_type
2692 - nir: Add alignment information to cast derefs
2693 - nir: Handle all array stride cases in nir_deref_instr_array_stride
2694 - nir: Add a helper for getting the alignment of a deref
2695 - nir/lower_io: Apply alignments from derefs when available
2696 - nir/opt_deref: Don't remove casts with alignment information
2697 - nir/opt_deref: Remove restrictive alignment information from casts
2700 - nir: Allow var_mem_global in nir_lower_vars_to_explicit_types
2701 - nir: Allow uniform in nir_lower_vars_to_explicit_types
2704 - clover/nir: Use lower_vars_to_explicit for uniform and global
2710 - nir/lower_io: Fix the unknown-array-index case in get_deref_align
2711 - nir: Add a dominance validation pass
2713 - intel/nir: Call validate_ssa_dominance at both ends of the NIR compile
2714 - nir: More NIR_MAX_VEC_COMPONENTS fixes
2715 - nir/idiv_const: Use the modern nir_src_as\_\* constant helpers
2720 - nir: Add a pass for lowering CL-style image ops to texture ops
2722 - nir: Rename get_buffer_size to get_ssbo_size
2725 - nir/liveness: Consider if uses in nir_ssa_defs_interfere
2728 - nir: Add a new memcpy intrinsic
2729 - nir: Add a lowering pass to lower memcpy
2731 - clover/nir: Call the memcpy lowering pass
2732 - nir: Allow creating variables with nir_var_mem_push_const.
2733 - nir/lower_io: Add support for push constants
2738 - nir/copy_propagate: Copy-prop into jump conditions
2739 - nir: Disallow goto and goto_if in clone and \[de]serialize
2740 - nir/cf: Better handle intra-block splits
2741 - nir/validate: Improve the validation of blocks
2742 - nir/lower_goto_ifs: Don't destroy SSA form in the process
2743 - nir/dominance: Use \_mesa_set_clear instead ofhand-rolling it
2745 - nir/lower_goto_ifs: Use rzalloc
2746 - nir/lower_goto_ifs: Add asserts for SSA forks
2747 - nir/lower_goto_ifs: Always include level dom_frontiers in prev_frontier
2748 - Revert "nir/lower_goto_if: Add a route::outside set"
2752 - nir: Split NIR_INTRINSIC_TYPE into separate src/dest indices
2753 - nir: Add a conversion and rounding intrinsic
2754 - nir: Add builder helpers for OpenCL type conversions
2755 - nir: Add a passes for nir_intrinsic_convert_alu_types
2759 - clover/nir: Call nir_lower_convert_alu_types
2760 - nir: Add lowering from regular ALU conversions to the intrinsic
2762 - nir: Fix a misspelling
2763 - nir/find_array_copies: Properly discard copies for casts
2764 - nir: Handle memcpy in copy_prop_vars and combine_stores
2765 - nir: Add a memcpy optimization pass
2766 - nir/opt_load_store_vectorize: Use bit sizes when checking mask compatibility
2767 - nir: Add component mask re-interpret helpers
2768 - nir/opt_deref: Add an instruction type switch
2769 - nir/opt_deref: Add an optimization for bitcasts
2770 - nir: Add a pass to lower vec3s to vec4s
2775 - nir/constant_folding: Use the builder
2776 - nir/constant_folding: Use nir_shader_instruction_pass
2777 - nir: Validate constant initializers
2778 - nir/constant_folding: Fold load_deref of nir_var_mem_constant
2787 - intel/nir: Don't try to emit vector load_scratch instructions
2788 - intel/nir: Lower load_global_constant in lower_mem_access_bit_sizes
2797 - clover: Stop leaking NIR shaders
2798 - nir/opt_deref: Fix the vector bitcast optimization
2799 - nir: Allow more deref modes in phis
2818 - nir/lower_io: Assert non-zero power-of-two alignments
2821 - clover/nir: Calculate sizes of images and samplers properly
2822 - clover/nir: Add an image lowering pass
2824 - nir/lower_memcpy: Don't mask the store
2826 - nir/validate: Explain why we don't use nir_foreach_block
2828 - nir/builder: Add a nir_ieq_imm helper
2829 - nir/phis_to_scalar: Use a deny-list for load_deref modes
2830 - nir: Handle incomplete derefs in split_struct_vars
2831 - nir: Use var->data.mode instead of deref->mode in a few cases
2832 - nir: Disallow writes to system values and mem_constant
2833 - nir/opt_find_array_copies: Allow copies from mem_constant
2834 - nir: Add and use some deref mode helpers
2835 - nir/lower_array_deref_of_vec: Use nir_deref_mode_must_be
2836 - nir/lower_io: Use nir_deref_mode\_\* helpers
2837 - nir/phis_to_scalar,gcm: Use nir_deref_mode_may_be
2838 - nir: Only force loop unrolling if we know it's a in/out/temp
2839 - nir/vars_to_ssa: Use nir_deref_must_be
2840 - nir/vec3_to_vec4: Use nir_deref_must_be
2841 - nir: Use nir_deref_mode_may_be in deref optimizations
2842 - nir/find_array_copies: Prepare for generic pointers
2843 - nir/split_*_vars: Prepare for generic pointers
2844 - nir: Make nir_deref_instr::mode a bitfield
2845 - nir: Add support for generic pointers
2847 - nir/opt_deref: Add a deref mode specialization optimization
2848 - nir/opt_deref: Add an optimization for deref_mode_is
2849 - nir/lower_io: Add a mode parameter to build_addr_iadd
2850 - nir/lower_io: Add a mode parameter to addr_format_is\_\*
2851 - nir/lower_io: Add support for 32/64bit_global for shared
2852 - nir/lower_io: Add support for lowering deref_mode_is
2853 - nir/lower_io: Support generic pointer access
2854 - nir/lower_io: Add a new 62bit_generic address format
2855 - nir/opt_intrinsics: Report progress for the gl_SampleMask optimization
2856 - nir/constant_folding: Use a switch in try_fold_intrinsic
2857 - nir/constant_folding: Use the standard variable naming convention
2858 - nir: Move constant folding of vote to opt_constant_folding
2859 - nir/constant_folding: Fold subgroup shuffle intrinsics
2860 - nir/opt_intrinsics: Refactor a bit
2861 - nir/opt_intrinsic: Optimize bcsel(b, shuffle(x, i), shuffle(x, j))
2862 - nir/find_array_copies: Don't assume all children exist
2863 - nir/deref: Fix a typo
2869 - spirv,nir: Add support for ray-tracing built-ins
2870 - nir/builder: Add a select_from_ssa_def_array helper
2871 - nir: Add intrinsics for object to/from world RT sysvals
2872 - nir: Add new variable modes for ray-tracing
2874 - nir,spirv: Add support for the ShaderCallKHR scope
2875 - spirv,nir: Add ray-tracing intrinsics
2876 - nir: Handle ray-tracing intrinsics and storage classes in copy-prop etc.
2878 - nir: Print formats on image intrinsics as text
2879 - nir: Validate image atomic formats
2882 - nir: Allow 64-bit image atomics
2884 - nir/lower_bit_size: Don't cast comparison results
2885 - nir/lower_bit_size: Pass a nir_instr to the callback
2886 - nir/lower_bit_size: Add support for lowering subgroup ops
2887 - intel/nir: Refactor lower_bit_size_callback
2888 - intel/nir: Lower 8-bit scan/reduce ops to 16-bit
2889 - intel/nir: Lower 8-bit ops to 16-bit in NIR on Gen11+
2895 - nir: nir_range_analysis needs to be updated for vec16
2899 - nir: Fix serialize/deserialize of void samplers/images
2900 - nir: Optimize mask+downcast to just downcast
2901 - nir: Add nir_address_format_32bit_offset_as_64bit
2902 - nir: Add nir_address_format_32bit_index_offset_pack64
2903 - nir/vtn: CL SPIR-V callers should specify address modes
2905 - nir: Add fisnormal op
2906 - nir/vtn: Support SpvOpIsNormal via fisnormal
2907 - nir: Add fisfinite op
2908 - nir/vtn: Support SpvOpIsFinite via fisfinite
2909 - nir/vtn: Handle LessOrGreater deprecated opcode
2910 - nir/vtn: Support OpOrdered and OpUnordered opcodes
2911 - nir/glsl: Add glsl_get_cl_type_size_align helper
2912 - nir: Use 'unsigned' instead of enum types in nir_variable::data
2914 - nir: Populate some places where existing system values were missing
2915 - nir: Add new system values and intrinsics for dealing with CL work offsets
2916 - nir: Move compute system value lowering to a separate pass
2917 - nir: Add options to nir_lower_compute_system_values to control compute ID base lowering
2919 - nir: Add a lowering pass to split 64bit phis
2920 - nir: Relax opt_if logic to prevent re-merging 64bit phis for loop headers
2922 - nir: Implement mul_high lowering for bit sizes other than 32
2923 - nir: Remove 32bit restriction for uadd_carry optimization
2924 - nir: Add bit_count to lower_int64 pass
2925 - nir/vtn: SPIR-V bit count opcodes (core and extension) dest size mismatches nir
2926 - clover/nir/spirv: Use uniform rather than shader_in for kernel inputs
2927 - nir/vtn: Add type constant to image intrinsics
2928 - nir/vtn: Add support for kernel images to SPIRV-to-NIR.
2929 - nir/vtn: Use return type rather than image type for tex ops
2930 - nir/vtn: Handle integer sampling coordinates
2931 - nir/vtn: ImageSizeLod op can be applied to images
2932 - nir/vtn: Add intrinsics for CL image format/order queries
2933 - nir/vtn: Convert constant samplers to variables with data
2935 - nir: More NIR_MAX_VEC_COMPONENTS fixes
2939 - nir: Add an internal flag to shader_info
2940 - nir: Add glsl_base_type unsigned -\> signed version helper
2941 - nir/vtn: Add handling for SPIR-V event variables
2950 - vtn/opencl: Switch some nir-sequence ops to use libclc
3047 - nir/algebraic: optimize iand/ior of (n)eq zero when umax/umin not available
3070 - nir: Add a structurizer
3076 - nir: Add a structured flag to nir_shader
3077 - nir: Add goto_if jump instruction
3080 - clover/nir: fix mem_shared by using address_format_32bit_offset
3081 - nv50/ir/nir: fix smem size
3082 - nv50/ir/nir: rework indirect function_temp handling
3083 - clover/nir: Call vars_to_explicit_types for shared memory
3085 - nv50/ir/nir: assert on unknown alu ops
3086 - clover/nir: support int64 atomics if the device supports it
3087 - nv50/ir/nir: fix global_atomic_comp_swap
3089 - nv50/ir/nir: support load_work_dim
3092 - nir: rename nir_op_fne to nir_op_fneu
3093 - nir: fix nir_variable_create for kernels
3094 - clover/nir: add support for global invocation id offsets
3097 - nir: use nir_var_all to get rid of casting
3099 - nir: use enum operator helper for nir_variable_mode and nir_metadata
3100 - clover/nir: Lower function_temp to scratch.
3102 - clover/nir: use offset for temp memory
3108 - nir/serialize: fix serialization of system values
3111 - clover/nir: extract constant buffer into its own section
3117 - clover/nir: set kernel_image cap
3121 - nv50/ir/nir: don't use designated initializers
3126 - nir: Copy semantics to nir_intrinsic_load_fs_input_interp_deltas
3127 - nir: Move new edgeflag assert into the io_lowered case
3130 - Revert "nir: replace lower_ffma and fuse_ffma with has_ffma"
3139 - nir/algebraic: Avoid creating new fp64 ops when using softfp64
3206 - blorp: identify copy kernels in NIR
3224 - gallium: Fix NIR validation when lowering polygon stipple
3274 - intel/compiler: use the same name for nir shaders in brw_compile\_\* functions
3303 - nir: handle float atomics in copy propagation pass
3321 - nir: remove nir_strip stub declaration
3322 - nir: handle load_input_vertex in nir_get_io_offset_src
3323 - nir: save IO semantics in lowered IO intrinsics
3324 - nir: gather all IO info from IO intrinsics
3325 - nir: update IO semantics in nir_io_add_const_offset_to_base
3326 - nir: print IO semantics (v2)
3327 - nir: properly identify texcoords for lowered IO in nir_lower_drawpixels
3328 - nir: add shader_info::io_lowered
3329 - nir: add interpolation qualifiers for color sysvals into shader_info
3330 - nir: generate lowered IO in nir_lower_passthrough_edgeflags
3331 - st/mesa: don't pass NIR to draw module if IO is lowered
3332 - st/mesa: don't generate NIR for ARB_vp/fp if NIR is not preferred
3343 - nir: fix a bug in is_dual_slot in nir_io_add_const_offset_to_base
3348 - ac/nir: handle all lowered IO intrinsics
3350 - radeonsi: remove in/out/uniform variables from NIR after lowering IO
3369 - nir: get ffma support from NIR options for nir_lower_flrp
3370 - nir/algebraic: trivially enable existing 32-bit patterns for all bit sizes
3371 - nir/algebraic: add 16-bit versions of a few 32-bit patterns
3421 - nir: add mediump flag to IO semantics
3422 - nir: fix lower_mediump_outputs to not require variables
3423 - nir/algebraic: add flrp patterns for 16 and 64 bits
3424 - nir/algebraic: expand existing 32-bit patterns to all bit sizes using loops
3425 - nir: remove redundant opcode u2ump
3426 - nir: enforce 32-bit src type requirement for f2fmp and i2imp
3427 - nir: add new mediump opcodes f2[ui]mp, i2fmp, u2fmp
3428 - nir/algebraic: collapse conversion opcodes (many patterns)
3429 - nir/algebraic: add late optimizations that optimize out mediump conversions (v3)
3430 - nir/opt_vectorize: don't lose exact and no_*_wrap flags
3431 - st/mesa: don't enable NV_copy_depth_to_color if NIR doesn't support FP64
3432 - nir,radeonsi: move ffma fusing to late optimizations for better codegen
3455 - nir: split fuse_ffma into fuse_ffma16/32/64
3456 - nir: split lower_ffma into lower_ffma16/32/64
3458 - nir: replace lower_ffma and fuse_ffma with has_ffma
3468 - nir: gather indirect info from lowered IO intrinsics
3469 - nir: gather tess.tcs_cross_invocation info from lowered IO intrinsics
3470 - nir: set system_values_read for all intrinsics
3471 - nir: gather fs.uses_sample_qualifier from lowered IO
3472 - nir: fix input/output info gathering for lowered IO
3473 - nir: gather information about fbfetch and dual source color
3475 - radeonsi: call nir_shader_gather_info after lowering and optimizing NIR
3479 - radeonsi: run NIR optimizations that glsl_to_nir runs but other places might not
3516 - nir/algebraic: always lower idiv to shifts if bitops are allowed
3518 - nir: add new helper passes that lower uniforms to literals
3538 - nir: consider load_color intrinsics as both inputs and sysval in gathering
3539 - Revert "st/mesa: don't pass NIR to draw module if IO is lowered"
3540 - st/mesa: make sure prog->info is up to date for NIR (v2)
3577 - nir: fix gathering TCS cross invocation access with lowered IO
3578 - nir: fix gathering patch IO usage with lowered IO
3579 - ac/nir: fix a typo in ac_are_tessfactors_def_in_all_invocs
3595 - scons: gallium/auxiliary: Unconditionally compile NIR regardless of LLVM
3781 - nir: update ubo locations in nir_lower_uniforms_to_ubo
3831 - nir/clip_disable: write 0s instead of undefs for disabled clip planes
3832 - nir/clip_disable: try for better no-op
3833 - nir/clip_disable: handle 2x vec4 case
3906 - clover/nir: Register callback for translation messages (v2)
3989 - nir/opt_remove_phis: optimize out phis with undef
3992 - nir: fix potential left shift of a negative value
3993 - nir: fix memory leak in nir_cf_list_clone
3998 - nir: add and use nir_intrinsic_has\_ helpers
4027 - nir/load_store_vectorize: rework alignment calculation
4028 - nir/opt_shrink_vectors: shrink image stores using the format
4031 - nir/opt_if: fix opt_if_merge when destination branch has a jump
4032 - nir/opt_loop_unroll: fix is_access_out_of_bounds with vectors
4046 - aco: use nir's constant source helpers more
4049 - st/nir: call nir_opt_access before gl_nir_lower_buffers
4051 - nir/instr_set: hash intrinsic sources
4052 - nir/load_store_vectorize: improve vectorization with identical operations
4057 - nir: return progress from nir_lower_io_to_scalar_early
4063 - android: fix SPIR-V -\> NIR build
4066 - radv/aco,nir/lower_subgroups: don't lower elect
4067 - nir: add last_invocation intrinsic
4069 - nir: move divergence analysis options to nir_shader_compiler_options
4070 - nir: allow divergence information to be updated when inserting instruction
4071 - nir: add pass to optimize uniform atomics
4073 - nir/opt_uniform_atomics: optimize image atomics
4074 - nir/opt_uniform_atomics: don't optimize atomics twice
4076 - scons: fix SPIR-V -\> NIR build
4077 - nir/opt_uniform_atomics: remove useless returns
4081 - nir/opt_load_store_vectorize: don't vectorize stores across demote
4082 - nir/opt_load_store_vectorize: add some tests for discard/demote behaviour
4085 - nir/loop_analyze: adjust force unrolling to only include interesting modes
4086 - ac/nir: remove bindless image atomic format check
4119 - nir: scalarize fdot in reverse
4121 - nir/algebraic: better propagate constants up fadd chains
4122 - nir: add nir_alu_src_is_trivial_ssa()
4123 - nir: skip bcsel with non-trivial swizzle in opt_simplify_bcsel_of_phi()
4124 - nir: use nir_alu_src_is_trivial_ssa() in nir_ssa_for_alu_src()
4125 - nir: add shader_info::bit_sizes_used
4126 - nir/lower_bit_size: optimize upcast of b2i8/b2i16
4136 - ac/nir: implement 64-bit images
4144 - nir/unsigned_upper_bound: fix buffer overflow in search_phi_bcsel
4145 - nir: fix sampler_lod_parameters_pan indices
4266 - radv: report the spirv-nir logs back to the application
4305 - nir/algebraic: mark some optimizations with fsat(NaN) as inexact
4328 - nir/lower_memory_model: return progress when visiting instructions
4329 - nir/lower_memory_model: do not break with global atomic operations
4330 - ac/nir: implement nir_intrinsic_{load,store}_global
4331 - ac/nir: implement nir_intrinsic_global_atomic\_\*
4334 - nir/lower_io: change nir_io_add_const_offset_to_base to use bitfield modes
4341 - ac/nir: fix nir_intrinsic_shared_atomic_fadd
4347 - ac/nir: remove dead load/store deref code for temporary variables
4357 - ac/nir: remove dead global load/store/atomic derefs code
4358 - ac/nir: remove dead shader IO code
4363 - ac/nir,radv: fix invalid IR when loading inline uniform blocks
4364 - nir/constant_folding: init nir_const_value to zero
4365 - aco: bail out if the NIR IO base offset isn't zero
4367 - ac/nir: implement nir_op_fsat
4372 - radv: use the same NIR compiler options for both compiler backends
4373 - radv/llvm: stop assigning driver_location in NIR->LLVM
4375 - ac/nir: pass the variable location to store_tcs_outputs
4376 - radv/llvm: switch to NIR IO assigned locations
4377 - radv/llvm: reduce the ESGS itemsize by using NIR IO assigned locations
4378 - radv/llvm: reduce LDS size for tess by using NIR IO assigned locations
4391 - aco: compute the CS workgroup size from the shader NIR info
4394 - ac/nir: implement missing nir_op_pack_half_2x16_split
4398 - ac/nir: handle non-const offset with txf/txf_ms
4399 - radv: move all NIR pass outside of ACO
4400 - ac/nir: do not sign-extend the result of texop_samples_identical
4403 - ac/nir: ignore set_vertex_and_primitive_count intrinsic
4404 - ac/nir: abort when an unknown intrinsic is reached
4416 - aco: store NIR range analysis data to the isel context
4418 - nir/algebraic: distribute imul(iadd(a, b), c) when b and c are constants
4420 - nir/algebraic: optimize bitfield_select(a, b, 0) to iand(a, b)
4455 - gallivm/nir: handle nir_op_flt in lp_build_nir_llvm
4488 - glsl: don't duplicate state vars as uniforms in the NIR linker
4504 - aco: Use NIR IO semantics for tess factor IO locations.
4506 - nir: Add ability to count emitted GS primitives.
4507 - nir: Add ability to count emitted GS vertices per primitive.
4508 - nir: Add ability to overwrite incomplete GS primitives.
4509 - nir: Count vertices per stream.
4510 - nir: Add ability to count primitives per stream.
4535 - nir: Emit set_vertex_and_primitive_count for inactive streams.
4568 - nir/lower_idiv: Port recent LLVM fixes to emit_udiv
4579 - nir: Fix undefined behavior due to signed integer multiplication overflows
4580 - nir: Fix unaligned pointer access
4672 - nir/large_constants: only search for constant duplicates