Lines Matching refs:aco

77 - \[aco\] problem compiling compute pipeline
95 - radv/aco: Vertex explosion on RPCS3
125 - radv, aco: dEQP-VK.glsl.atomic_operations.*_fragment_reference regressed
874 - aco: Add VK_KHR_shader_terminate_invocation support.
1178 - aco: execute branch instructions in WQM if necessary
1180 - aco/isel: refactor code and remove unnecessary v_mov
1181 - aco/isel: refactor emit_vop3a_instruction() to handle 2 operand instructions
1183 - aco: propagate SGPRs into VOP1 instructions early.
1184 - aco: expand create_vector more carefully w.r.t. subdword operands
1185 - aco: use p_create_vector for nir_op_pack_half_2x16
1187 - aco: use p_split_vector for nir_op_unpack_half_\*
1188 - aco: add validation rules for p_split_vector
1189 - aco: use v_cvt_pkrtz_f16_f32 for pack_half_2x16
1190 - radv,aco: lower_pack_half_2x16
1191 - aco: use VOP2 version of v_cvt_pkrtz_f16_f32 on GFX_6_7_10
1192 - aco: use VOP2 for v_cvt_pkrtz_f16_f32 if possible
1193 - aco: refactor GFX6_7 subdword copy lowering
1194 - aco: improve code sequences for 16bit packing
1195 - aco: use do_pack() for self-intersecting operations.
1196 - aco: fix GFX8 16-bit packing
1197 - aco: implement nir_op_unpack_[64/32]_\*
1202 - aco: refactor split_store_data() to always split into evenly sized elements
2575 - util,ac,aco,radv: Cross-platform memstream API
2578 - aco: Clean up some C++ usages
2580 - aco: Fix accidental copies, attempt two
2582 - amd/llvm,aco: Replace VLA with alloca
3648 - android: aco/isel: Move context initialization code to a dedicated file
3659 - android: util,ac,aco,radv: Cross-platform memstream API
3661 - android: aco: add aco_form_hard_clauses.cpp to Makefile.sources
3986 - aco: fix C++11/C++14 compilation
3987 - aco: set constant_data_offset correctly in the case of merged shaders
3988 - aco: don't move memory accesses to before control barriers
3997 - aco: fix non-rtz pack_half_2x16
3999 - aco: use nir_intrinsic_has_access
4001 - aco: consider branch definitions in spiller
4002 - aco: don't consider the first partial spill if it's the wrong type
4003 - aco: don't fix break condition for break+discard to exec
4004 - aco: fix regclass checks when fixing to vcc/exec with Builder
4005 - aco: fix spills_entry heuristic for branch blocks in init_live_in_vars()
4006 - aco: keep loop live-through variables spilled
4007 - aco: reserve 2 sgprs for each branch
4008 - aco: create long jumps
4009 - aco/tests: add test for GFX10 0x3f bug
4010 - aco: shorten disassembly for repeated instructions
4011 - aco/tests: add tests for long jumps
4012 - aco: remove 64-bit SGPR ubfe/ibfe
4013 - aco: fix sgpr ubfe/ibfe if the offset is too large
4014 - aco: sink get_alu_src() in bfe lowering
4024 - aco: fix mad splitting after applying output modifiers
4025 - aco: remove omod_success/clamp_success
4026 - aco: fix byte_align_scalar for 3 dword vectors
4029 - aco: fix one-off error in Operand(uint16_t)
4030 - aco: improve fsign selection
4033 - aco: fix v_writelane_b32 with two sgprs
4034 - aco: workaround disassembler bug of v_writelane_b32 with literal
4035 - aco: don't apply constant to SDWA on GFX8
4036 - aco: fix value numbering of reductions
4037 - aco: fix validation of sub-dword parallel-copies
4038 - aco: pass -fno-exceptions and -fno-rtti
4039 - aco: fix incorrect assertion in emit_vop3a_instruction()
4041 - radv,aco: fix reading primitive ID in FS after TES
4042 - aco: keep track of temporaries' regclasses in the Program
4043 - aco: use bit vectors for liveness sets
4044 - aco: use io semantics to get an intrinsic's slot
4045 - aco: use nir_get_io_offset_src() in visit_load_input()
4046 - aco: use nir's constant source helpers more
4047 - aco: remove dead indirect fs input loading
4048 - aco: stop multiplying driver_location by 4
4053 - aco: fix get_buffer_resource_flags()
4054 - aco: remove trailing whitespace
4056 - aco: Add loop creation helpers.
4064 - aco: optimize more uniform reductions/scans
4065 - aco: implement elect
4066 - radv/aco,nir/lower_subgroups: don't lower elect
4068 - aco: implement last_invocation
4072 - aco: use nir_opt_uniform_atomics
4075 - aco: fix get_ssbo_size with a vgpr resource
4078 - aco: implement 16-bit literals
4079 - aco: propagate literals into sub-dword pseudo instructions on GFX9+
4080 - aco: don't use v_pack_b32_f16 if 16-bit input denormals are flushed
4083 - aco: add missing SCC clobber in get_buffer_size
4087 - aco: remove isel_context::allocated
4088 - aco: update phi_map in add_subdword_operand()
4089 - aco: don't do divergent break+discard
4090 - aco: skip value numbering of copies
4091 - aco: copy-propgate through p_create_vector during value numbering
4092 - aco: expand vectors passed as copy operands
4093 - aco: don't use bld.copy() in handle_operands()
4094 - aco: allow literals on sub-dword p_parallelcopy
4095 - aco: always use p_parallelcopy for pre-RA copies
4096 - aco: use Builder::copy more
4097 - aco: remove some unused optimizations
4098 - aco: use v_mov_b32_sdwa for some 16-bit constants
4099 - aco: remove all-undef phi opt
4100 - aco: ignore the ACO-inserted continue in create_continue_phis()
4101 - aco: default to a definition size of 32
4102 - aco: round bytes_written to dwords if larger than 4 bytes
4103 - aco: use control flow creation helpers in select_gs_copy_shader
4104 - aco: use mubuf helper in select_gs_copy_shader
4105 - aco: move individual instruction disassembly to its own helper
4106 - aco: refactor repeated instruction disassembly
4107 - aco: switch aco_print_asm to a FILE \\*
4108 - aco: create s_clause on GFX10+
4109 - aco: assert a label only uses one of the members in ssa_info's union
4110 - aco: fix printing of some sdwa sels
4111 - aco: fix combine_inverse_comparison()
4112 - aco: don't allow destination opsel for v_cvt_pknorm
4113 - aco: handle SDWA in the optimizer
4130 - aco: implement some 16-bit arithmetic instead of lowering
4131 - aco: implement 8/16-bit instructions which can be trivially widened
4135 - aco: implement 64-bit images
4138 - aco: don't combine precise max(min()) to med3
4139 - aco: fix combine_constant_comparison_ordering() NaN check with 16/64-bit
4140 - aco: disallow various v_add_u32 opts if modifiers are used
4141 - aco: disable omod if the sign of zeros should be preserved
4142 - aco: fix fp16 \*0.5 omod
4143 - aco: fix v_mul_hi_u32_u24 format
4275 - aco: do not set valid_mask for POS0 exports on GFX 10.3
4277 - aco: rename DEBUG_VALIDATE to DEBUG_VALIDATE_IR
4278 - aco: rework the way various compilation/validation errors are reported
4279 - radv,aco: report ACO errors/warnings back via VK_EXT_debug_report
4280 - aco: fix file leak in ra_fail()
4282 - aco: add ACO_DEBUG=force-waitcnt to emit wait-states
4285 - aco: add TBA/TMA/TTMP0-11 physical registers definitions
4286 - aco: validate that SMEM operands can use fixed registers
4287 - aco: add a helper for building a trap handler shader
4288 - aco: skip unnecessary compiler pass for the trap handler program
4297 - aco: add ACO_DEBUG=novn,noopt,nosched for debugging purposes
4300 - aco: fix wrong source position for constant with nir_op_cube_face_coord
4307 - radv,aco: disable opts if VK_PIPELINE_CREATE_DISABLE_OPTIMIZATION_BIT
4308 - aco: handle unaligned loads on GFX10.3
4343 - radv/aco: lower IO for all stages outside of ACO
4344 - aco: apply the clamped integer addition disassembly workaround for v_add3
4345 - aco/tests: add disassembler tests to reproduce the add3+clamp crash
4365 - aco: bail out if the NIR IO base offset isn't zero
4366 - aco: more uses of nir_get_io_offset_src()
4370 - aco: dump the program if the disassembler failed
4384 - aco: implement missing nir_op_unpack_half_2x16_split_{x,y}_flush_to_zero
4385 - radv/aco: disable NGG GS support because it randomly hangs the GPU
4387 - aco: remove stub lower_wqm() prototype
4388 - aco: remove useless occurences of radv_nir_compiler_options
4389 - aco: remove unused radv_shader.h includes
4391 - aco: compute the CS workgroup size from the shader NIR info
4392 - aco: adjust an assertion about the wavesize in emit_gfx10_wave64_bpermute()
4397 - Revert "radv/aco: disable NGG GS support because it randomly hangs the GPU"
4401 - radv,aco: fix use of texop_samples_identical in the resolve meta path
4402 - aco: fix determining if LOD is zero for nir_texop_txf/nir_texop_txs
4414 - radv,aco: adjust the sample mask only if per-sample shading is enabled
4415 - radv,aco: optimize computing the sample mask for per-sample shading
4416 - aco: store NIR range analysis data to the isel context
4417 - aco: select v_mul_{hi}_u32_u24 for 24-bit multiplications
4419 - aco: optimize v_and(a, v_subbrev_co(0, 0, vcc)) -\> v_cndmask(0, a, vcc)
4421 - aco: fix combining add/sub to b2i if a new dest needs to be allocated
4496 - aco: Fix unused variable warning by adding ASSERTED.
4497 - aco: Fix convert_to_SDWA when instruction has 3 operands.
4498 - aco: Move README to README-ISA
4499 - aco: Fixup markdown formatting of the README-ISA.
4500 - aco: Add README which explains about what ACO is and how it works.
4501 - aco: Fix emit_boolean_exclusive_scan in wave32 mode.
4502 - aco: Clean up emit_mbcnt.
4503 - aco: Add base argument to emit_mbcnt.
4504 - aco: Use NIR IO semantics for tess factor IO locations.
4505 - radv/aco: Set I/O variable locations outside ACO.
4511 - radv/aco: Use new GS lowering options for ACO with NGG GS.
4512 - aco: Clarify missing export error message in assembler.
4513 - aco: Extract lanecount_to_mask to a separate function.
4514 - aco: Extract thread_id_in_threadgroup to a separate function.
4515 - aco: Use thread_id_in_threadgroup helper for ES outputs.
4516 - aco: Optimize thread_id_in_threadgroup when there is just one wave.
4517 - aco: Add wave-specific opcode for s_lshl and s_flbit.
4518 - aco/ngg: Refactor gs_alloc_req in preparation for NGG GS.
4519 - aco/ngg: Refactor ngg_emit_prim_export in preparation for NGG GS.
4520 - aco/ngg: Make primitive export packing less prone to error.
4521 - aco/ngg: Clean up and reorganize NGG VS/TES code.
4522 - aco/ngg: Allow NGG GS to store ES outputs.
4523 - aco/ngg: Allow NGG GS to load per-vertex GS inputs.
4524 - aco/ngg: Allow NGG GS to create VS exports.
4525 - aco/ngg: Setup NGG GS.
4526 - aco/ngg: Create LDS layout for NGG GS.
4527 - aco/ngg: Implement workgroup reduce / exclusive scan for NGG GS.
4528 - aco/ngg: Implement NGG GS output.
4529 - aco/ngg: Place workgroup barrier outside control flow for NGG GS.
4530 - aco/ngg: Add shader query support to NGG GS.
4531 - radv/aco: Enable NGG GS by default.
4532 - aco/ngg: Use more efficient LDS layout to help reduce bank conflicts.
4533 - aco/ngg: Allocate NGG GS space early for const vertex/primitive counts.
4534 - aco/ngg: Calculate workgroup size of NGG shaders.
4536 - aco/ngg: Add assertion to make sure we always know the vertex count.
4537 - aco: Assert that workgroup barriers are not used inappropriately.
4538 - aco/ngg: Put shader query reduction operand into a VGPR.
4539 - aco: Add some validation for PSEUDO_REDUCTION instructions.
4540 - aco: Make emitting reduction instructions a bit more convenient.
4541 - aco: Add a few assertions about LDS usage.
4542 - aco/ngg: Export a zero-area triangle when primitive count is 0.
4543 - aco/ngg: Incorporate GS invocations into workgroup size calculation.
4544 - aco/optimizer: Only set scc_needed when it is actually needed.
4545 - aco: Fix NGG GS assert failure from the WG scan.
4570 - aco: Fix integer overflows when emitting parallel copies during RA
4572 - aco/isel: Turn the function template emit_load into a proper function
4573 - aco/isel: Simplify nested branching code
4574 - aco/isel: Consistently use references for input parameters in emit_load
4575 - aco/isel: Remove unused definitions
4576 - aco/isel: Move context initialization code to a dedicated file
4577 - aco/isel: Move add_startpgm to aco_instruction_selection.cpp
4578 - aco/isel: Compile all helper functions with static linkage
4585 - aco/isel: Fix out-of-bounds write in visit_load_input
4586 - aco/isel: Always export position data from VS/NGG
4587 - aco/isel: Remove some dead code
4588 - aco/isel: Remove now unused VS-related code from create_null_export
4589 - aco: Use strong typing to model SW<->HW stage mappings
4590 - aco: Clean up symbol names and comments related to NGG
4591 - aco/isel: Miscellaneous cleanups using the new Stage API
4592 - aco/ra: Fix counting of subdword variables in get_reg_create_vector
4593 - aco: Fix format string used when raising validation errors
4625 - aco: Initialize mad_info member literal_idx.