aboutsummaryrefslogtreecommitdiffstats
path: root/src/amd/compiler/aco_optimizer.cpp
Commit message (Expand)AuthorAgeFilesLines
* aco: shrink mad_infoRhys Perry2020-06-181-1/+1
* aco: make ssa_info::label 64-bitRhys Perry2020-06-181-17/+20
* aco: shrink ssa_infoRhys Perry2020-06-181-1/+1
* aco: add more opcodes to can_swap_operandsRhys Perry2020-06-151-36/+53
* aco: allow to swap operands for some 16-bit float instructionsSamuel Pitoiset2020-06-151-0/+15
* aco: optimize 16-bit and 64-bit float comparisonsRhys Perry2020-06-151-36/+101
* aco: copy-propagate constants through p_extract_vector/p_split_vectorRhys Perry2020-06-151-5/+24
* aco: improve 8/16-bit constantsRhys Perry2020-06-151-81/+113
* aco: improve sub-dword check for sgpr/constant propagationRhys Perry2020-06-151-4/+9
* aco: create 16-bit input and output modifiersRhys Perry2020-06-151-22/+52
* aco: create 16-bit mad/fmaRhys Perry2020-06-151-6/+22
* aco: try to use fma instead of mad when denormals are enabledRhys Perry2020-06-151-17/+34
* aco: create mads when signed zeros should be preservedRhys Perry2020-06-151-1/+1
* aco: use num_opcodes instead of last_opcodeRhys Perry2020-06-151-4/+4
* aco: allow reading/writing upper halves/bytes when possibleRhys Perry2020-06-101-1/+1
* aco: use v_xor3_b32Rhys Perry2020-06-081-0/+3
* aco: Don't std::move temporary object.Oschowa2020-06-021-1/+1
* aco: preserve more fields when combining additions into SMEMRhys Perry2020-05-281-0/+4
* aco: fix missing break in label_instruction()Samuel Pitoiset2020-05-211-0/+1
* aco: optimize add/sub(a, cndmask(b, 0, 1, cond)) -> addc/subbrev_co(0, a, b)Samuel Pitoiset2020-05-121-13/+79
* aco: either copy-propagate or inline create_vector operandsDaniel Schürmann2020-05-071-4/+6
* aco: handle undef p_create_vector operands in the optimizerRhys Perry2020-04-241-0/+4
* aco: fix outdated label_vec from p_create_vector labellingRhys Perry2020-04-241-3/+5
* aco: fix v_or(s_lshl) and v_add(s_lshl) optimizationsRhys Perry2020-04-241-2/+2
* aco: combine VALU and SALU into various VOP3 instructionsRhys Perry2020-04-231-2/+9
* aco: copy-propagate p_create_vector copies of vectorsRhys Perry2020-04-231-4/+5
* aco: improve vector optimization with sub-dword vectorsRhys Perry2020-04-141-11/+22
* aco: fix p_extract_vector optimization in presence of unequally sized vector ...Daniel Schürmann2020-04-131-22/+27
* aco: don't constant-propagate into subdword PSEUDO instructionsDaniel Schürmann2020-04-101-6/+8
* aco: zero-initialize TempRhys Perry2020-04-061-0/+2
* aco: don't assume split_vector(create_vector) has the same number of elements...Daniel Schürmann2020-04-031-1/+2
* aco: don't propagate SGPRs into subdword PSEUDO instructionsDaniel Schürmann2020-04-031-2/+6
* aco: always optimize v_mad to v_madak in presence of literalsSamuel Pitoiset2020-04-031-2/+11
* aco: Fix combining DS additions in the optimizer.Timur Kristóf2020-03-111-5/+8
* aco: handle v_add_co_u32_e64 in parse_base_offset()Rhys Perry2020-03-031-0/+2
* aco: improve SCC handling in some SALU combinesRhys Perry2020-02-121-6/+4
* aco: disable some instruction combining if it could change an exec operandRhys Perry2020-02-121-2/+13
* aco: fix creating v_madak if v_mad_f32 has two sgpr literalsSamuel Pitoiset2020-02-111-1/+1
* aco/optimizer: Don't combine uniform bool s_and to s_andn2.Timur Kristóf2020-02-051-0/+3
* aco: reorder VMEM operands in ACO IRDaniel Schürmann2020-01-291-15/+10
* aco: fix combine_salu_not_bitwise() when SCC is usedDaniel Schürmann2020-01-281-0/+1
* aco: fix literal application with v_cndmask_b32/v_addc_co_u32/etcRhys Perry2020-01-271-0/+3
* aco: always add sgprs to sgpr_ids when choosing literalsRhys Perry2020-01-271-3/+5
* aco: Transform uniform bitwise instructions to 32-bit if possible.Timur Kristóf2020-01-241-0/+87
* aco: Make a better guess at which instructions need the VCC hint.Timur Kristóf2020-01-241-0/+16
* aco: fix constant folding of SMRD instructions on GFX6Samuel Pitoiset2020-01-201-1/+3
* aco: fix stack buffer overflow in apply_sgprs()Rhys Perry2020-01-201-1/+2
* aco: do not combine additions of DS instructions on GFX6Samuel Pitoiset2020-01-161-1/+5
* aco: Flip s_cbranch / s_cselect to optimize out an s_not if possible.Timur Kristóf2020-01-141-3/+44
* aco: Optimize out s_and with exec, when used on uniform bitwise values.Timur Kristóf2020-01-141-4/+31
generated by cgit v1.2.3 (git 2.39.1) at 2024-12-04 22:19:52 +0000