Lines Matching full:aco

71 - aco: Assertion when compiling CP2077 shader
93 - aco, radv Rage 2 menu corruption - bisected
94 - radv, aco: World War Z  character texture regression on 7900xtx
99 - Baldurs Gate 3 (DX11) - Graphical corruption on RDNA3 (ACO regression)
158 …ith constant 31 removed on width argument to BitFieldSExtract, causing incorrect result on RADV ACO
166 - radv,aco: In the game Rise of the Tomb Raider on RDNA 3 GPUs appeared artifacts after commit 290c…
177 - aco: KHR-GL46.shader_image_load_store.basic-allTargets-atomic asserts on Bonaire
217 - regression in aco,ac/llvm,radv,radeonsi: handle ps bc optimization in nir for radv
230 - [ACO] [RADV] Flickering squares in some areas in The Last of Us Part 1 (with workaround)
246 - aco: s_load_dword with negative soffset cause GPU hang
368 - aco,radv: Use unified atomics
501 - aco: Drop NIR parallel copy handling
686 - aco: fix nir_op_vec8/16 with 16-bit elements.
687 - aco: Fix some constant patterns in 16-bit vec4 construction with s_pack.
820 - aco: fix alignment check in emit_load
987 - aco: implement nir_intrinsic_load_resume_shader_address_amd
988 - aco: implement select_program_rt()
1003 - aco: adjust RT prolog for shader functions [disables RT]
1009 - aco/assembler: align resume shaders with cache lines
1010 - aco/assembler: align loops if it reduces the number of cache lines
1011 - aco/assembler: change prefetch mode on GFX10.3+ during loops if beneficial
1015 - amd: move end-of-code marker padding to ACO.
1017 - aco/insert_exec_mask: set Exact mode after p_discard_if when necessary
1670 - radv,aco: tweaks to get clang-format to print nicer code
1672 - aco: reformat according to its .clang-format
1673 - ci: enforce formatting for RADV & ACO
1786 - aco: use c++17
1970 - aco: Lower divergent bool phis iteratively
1972 - aco: Reset scratch_rsrc on blocks without predecessors
1973 - aco: Fix live_var_analysis assert
1974 - aco: Fix assert in insert_exec_mask
2014 - aco: Assert that operands have the same byte offset when reassigning split vectors
2015 - aco: also reassign p_extract_vector post ra
2016 - aco/vn: compare all valu modifers
2017 - aco/optimizer: don't use pass_flags for mad idx
2018 - aco/optimizer: copy pass flags for newly created valu instructions
2019 - aco/assembler: support VOP3P with DPP
2020 - aco/builder: support VOP3(P) with dpp
2021 - aco: add assembler tests for VOP3(P) with DPP
2022 - aco/ra: convert VOPC_DPP instructions without vcc to VOP3
2023 - aco: use VOP3+DPP
2024 - aco: don't apply dpp if the alu instr uses the operand twice
2025 - aco: emit_wqm on MIMG dst, not operands
2026 - aco: introduce helper to swap valu operands with modifiers
2027 - aco/gfx11: use fmamk/fmaak with opsel
2028 - aco: add withoutVOP3 helper
2029 - aco/ra: use smaller operand stride for VOP3P with DPP
2030 - aco/ra: use fmac with DPP/opsel on GFX11
2031 - aco: add helper function for can_use_input_modifiers
2032 - aco: use get_operand_size for dpp opt
2033 - aco: use can_use_input_modifiers helper
2034 - aco/optimizer: allow DPP to use VOP3 on GFX11
2037 - aco: cleanup v_cmp_class usage
2038 - aco: p_start_linear_vgpr doesn't always need exec mask
2039 - aco/ir: return true in hasRegClass for Operand(reg, rc)
2040 - aco/statistics: improve v_fma_mix dual issuing detection
2041 - aco: use v_add_f{16,32} with clamp for fsat
2042 - aco: use v_fma_mix for f2f32 and f2f16 on gfx11 if wave64
2043 - aco: make validation work without SSA temps
2044 - aco: move cfg validation to its own function
2045 - aco: don't validate p_constaddr_addlo/p_resumeaddr_addlo operands
2046 - aco: validate ir for prologs and after lower_to_hw_instr
2047 - aco/opcodes: move v_cndmask_b32 back to the VOP2 list
2048 - aco: remove v_cvt_pkrtz_f16_f32_e64 when it's actually VOP2
2049 - aco/opcodes: delete wrong comment copy pasted from NIR
2050 - aco: use uses helpers for pk_fma opt
2051 - aco: combine scalar mul+pk_add to pk_fma
2052 - aco/gfx10+: use v_cndmask with literal for reduction identity
2055 - aco: implement nir_op_bitz/bitnz
2057 - radv: set has_bit_test for aco
2058 - aco/optimizer: delete s_bitcmp optimization
2059 - aco/gfx11: fix get_gfx11_true16_mask with v_cmp_class_f16
2060 - aco: fix non constant 16bit bitnz/bitz
2061 - aco: fix u2f16 with 32bit input
3010 - aco/rt: Do not initialize the next shader addr
3108 - aco/spill: Make sure that offset stays in bounds
3462 - aco: don't treat ACCESS_NON_READABLE as ACCESS_COHERENT
4332 - aco: support 32bit address in nir_load_smem_amd
4334 - aco: implement nir_bindless_image_atomic_inc/dec_wrap
4335 - aco: skip scratch buffer init when its arg is not used
4336 - aco: fix nir_f2u64 translation
4338 - aco: move create_fs_dual_src_export_gfx11 above
4339 - aco: implement nir_export_dual_src_blend_amd
4340 - ac/nir/ps: use nir_export_dual_src_blend_amd when aco
4342 - aco: support nir_export_amd with ps targets
4343 - aco,radv: lower outputs to exports when nir for monolithic ps
4347 - aco,ac/llvm,radv,radeonsi: handle ps bc optimization in nir for radv
4348 - aco,radv: remove unused aco compile options
4349 - aco,radv: support symbol relocation in aco
4350 - aco: get scratch addr from symbol for radeonsi
4351 - aco: allow no export instruction for gfx10+ fs
4353 - meson: build radeonsi with aco
4354 - radeonsi: add aco debug option
4359 - radeonsi: init spi ps input shader config when aco
4370 - ac/llvm,radeonsi: lower some pack/unpack ops not supported by aco
4371 - ac/llvm,radeonsi: lower nir_fpow for aco and llvm
4372 - radeonsi: lower some 64bit ops aco does not support
4373 - radeonsi: lower vector const to scalar at last for aco
4375 - radeonsi: lower non uniform texture access when aco
4376 - radeonsi: add initial aco compile code
4378 - radeonsi: resolve aco scratch addr symbols
4379 - radeonsi: adjust ps args for aco
4381 - radeonsi: clamp shadow texture reference in nir for aco
4384 - radeonsi: be able to use aco compiler for mono ps
4386 - radeonsi: fix aco compile for atomic ops
4389 - aco: implement two load lds ngg intrininsic for radeonsi
4390 - aco,radv: remove unused aco_shader_info fields
4392 - aco: implement load buffer with ACCESS_USES_FORMAT_AMD
4393 - aco/assembler: handle ds_(add|sub)_gs_reg_rtn encoding
4394 - aco: use gds reg when ordered xfb counter add
4395 - aco: implement nir_xfb_counter_sub_amd
4396 - aco: implement nir_bindless_image_fragment_mask_load_amd
4397 - aco: use ac_get_image_dim for array check when image intrinsic
4398 - radeonsi: resolve lds ngg aco symbols
4399 - radeonsi: add scratch offset vs args explicitly for aco
4408 - radeonsi: fill aco shader info for mono standalone vs
4413 - radeonsi: enable aco for mono standalone vs
4414 - aco,radv: remove unused gs aco shader info
4417 - radeonsi: add scratch_offset arg for aco tcs
4418 - radeonsi: lower nir_load_tess_rel_patch_id_amd in abi for aco
4420 - radeonsi: enable aco support for mono standalone tcs
4421 - radeonsi: add scratch_offset arg for aco tes
4422 - radeonsi: init tes aco shader info fields
4424 - radeonsi: enable aco support for standalone tes
4425 - radeonsi: add scratch_offset arg for aco gs
4427 - radeonsi: enable aco for standalone gs
4428 - radeonsi: enable aco support for gs copy shader
4429 - radeonsi: add scratch_offset arg for aco cs
4432 - radeonsi: enable aco support for compute shader
4455 - aco: support implicit LOD for nir_texop_tg4
4457 - aco: remove SMEM_instruction::prevent_overflow
4458 - aco: use apply_nuw_to_ssa() with load_smem_amd
4460 - aco: don't move exec reads around exec writes
4461 - aco: don't move exec writes around exec writes
4463 - aco/ra: create M0-affinities for s_sendmsg
4464 - aco/gfx11: fix VMEM/DS->VALU WaW/RaW hazard
4469 - aco: add get_op_fixed_to_def() helper
4470 - aco: consider how definitions fixed to operands can change register demand
4472 - nir,vtn,aco,ac/llvm: make cube_face_coord_amd more direct
4478 - aco: remove unused RegType
4479 - aco: let p_start_linear_vgpr take an operand
4480 - aco: add MIMG_instruction::strict_wqm
4481 - aco: implement strict_wqm_coord_amd
4482 - aco: implement texture samples with strict WQM coordinates
4484 - aco/tests: improve performance of declaration parsing
4485 - aco/tests: add fix_derivs_in_divergent_cf tests
4486 - aco: fix update_alu(clear=true) for exports
4487 - aco: use pass_flags to recover s_delay_alu cycles
4488 - aco: insert s_delay_alu on the linear CFG
4489 - aco: improve printing of s_delay_alu
4491 - aco: fix has_color_exports=true for mrtz exports
4492 - aco/tests: add discard export target tests
4493 - aco: fix ds_sub_gs_reg_rtn validation
4496 - aco: consider position/primitive exports around memory barriers
4498 - aco: remove memory_barrier_buffer implementation
4499 - aco: mask bits source of s_bfe
4500 - aco/tests: test that s_bfe bits is masked
4503 - aco: run nir_lower_int64 after nir_opt_uniform_atomics
4507 - aco: don't try to form load+store clauses
4508 - aco/gfx11: use s_clause with stores
4509 - aco/gfx11: schedule for VMEM store clauses
4510 - aco: don't set exec_hi for wave32 scan reductions
4513 - aco: include helpers in emit_uniform_{reduce,scan}
4514 - nir,aco: add INCLUDE_HELPERS index to reduce intrinsic
4517 - aco,ac/llvm,ac/nir,vtn: unify cube opcodes
4522 - aco: remove 64-bit integer conversion opcodes
4531 - aco: insert s_nop before VGPR deallocation
4534 - aco: summarize register demand after handling branches
4535 - aco: don't create sendmsg(dealloc_vgprs) if scratch is used
4537 - aco: fix p_bpermute_gfx6 with input at non-zero byte
4539 - aco/spill: skip p_branch in process_block
4540 - aco/spill: add all live-in to merge block spill candidates
4541 - aco/optimizer_postRA: check overwritten_subdword in is_overwritten_since()
4542 - aco: check logical_phi_info at p_logical_end when eliminating exec writes
4543 - aco: remove unused p_logical_end check when optimizing branching sequence
4544 - aco: reset prefetch in the correct block after removing the exit
4846 - aco: remove nir_intrinsic_load_barycentric_at_sample occurences
5215 - aco: Disallow constant propagation on SOPP and fixed operands.
5224 - aco: Don't allow any VALU instruction to write m0.
5225 - aco: Initialize vcmpx field in get_cmp_info.
5243 - aco: Mark exec write used when it writes other registers.
5247 - aco: Use ac_hw_stage instead of aco-specific HWStage.
5248 - aco: Add hw_stage field to aco_shader_info.
5251 - aco: Use aco_shader_info::hw_stage instead of guessing.
5252 - aco: Remove unneeded stage related info fields.
5260 - aco: Add MESA_SHADER_KERNEL to instruction selection setup.
5261 - aco: Fix subgroup_id intrinsic on GFX10.3+.
5270 - aco/spill: Use arena allocator for next use distances
5271 - aco/spill: Use arena allocator for spills
5302 - aco: Support pops_exiting_wave_id PhysReg usage
5304 - aco: Add s_wait_event argument bit definitions
5305 - aco: Add Primitive Ordered Pixel Shading pseudo-instructions
5306 - aco: Skip waitcnt insertion in the discard early exit block
5307 - aco: Add Primitive Ordered Pixel Shading scheduling rules
5308 - aco: Send MSG_ORDERED_PS_DONE where necessary
5309 - aco: Add Primitive Ordered Pixel Shading waitcnt rules
5310 - aco: Implement fragment shader interlock intrinsics