projects
/
platform
/
upstream
/
mesa.git
/ history
commit
grep
author
committer
pickaxe
?
search:
re
summary
|
shortlog
|
log
|
commit
|
commitdiff
|
tree
first ⋅ prev ⋅
next
aco: combine scalar mul+pk_add to pk_fma
[platform/upstream/mesa.git]
/
src
/
amd
/
compiler
/
aco_optimizer.cpp
2023-06-20
Georg Lehmann
aco: combine scalar mul+pk_add to pk_fma
blob
|
commitdiff
2023-06-20
Georg Lehmann
aco: use uses helpers for pk_fma opt
blob
|
commitdiff
|
diff to current
2023-06-16
Eric Engestrom
aco: reformat according to its .clang-format
blob
|
commitdiff
|
diff to current
2023-06-07
Georg Lehmann
aco: use v_fma_mix for f2f32 and f2f16 on gfx11 if...
blob
|
commitdiff
|
diff to current
2023-06-07
Georg Lehmann
aco: use v_add_f{16,32} with clamp for fsat
blob
|
commitdiff
|
diff to current
2023-05-18
Georg Lehmann
aco/optimizer: allow DPP to use VOP3 on GFX11
blob
|
commitdiff
|
diff to current
2023-05-18
Georg Lehmann
aco: use can_use_input_modifiers helper
blob
|
commitdiff
|
diff to current
2023-05-18
Georg Lehmann
aco: use get_operand_size for dpp opt
blob
|
commitdiff
|
diff to current
2023-05-18
Georg Lehmann
aco/gfx11: use fmamk/fmaak with opsel
blob
|
commitdiff
|
diff to current
2023-05-18
Georg Lehmann
aco: introduce helper to swap valu operands with modifiers
blob
|
commitdiff
|
diff to current
2023-05-12
Georg Lehmann
aco: don't apply dpp if the alu instr uses the operand...
blob
|
commitdiff
|
diff to current
2023-05-12
Georg Lehmann
aco: use VOP3+DPP
blob
|
commitdiff
|
diff to current
2023-05-12
Georg Lehmann
aco/optimizer: copy pass flags for newly created valu...
blob
|
commitdiff
|
diff to current
2023-05-12
Georg Lehmann
aco/optimizer: don't use pass_flags for mad idx
blob
|
commitdiff
|
diff to current
2023-05-04
Timur Kristóf
aco: Disallow constant propagation on SOPP and fixed...
blob
|
commitdiff
|
diff to current
2023-04-19
Rhys Perry
aco: remove SMEM_instruction::prevent_overflow
blob
|
commitdiff
|
diff to current
2023-04-13
Harri Nieminen
amd: fix typos
blob
|
commitdiff
|
diff to current
2023-04-10
Timur Kristóf
aco: Fix optimization of v_cmp with subgroup invocation.
blob
|
commitdiff
|
diff to current
2023-03-31
Rhys Perry
aco: don't optimize s_or_b64(v_cmp_u_f32(a, b), cmp...
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco/optimizer: remove to_SDWA
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: keep label_mul/usedef/minmax in apply_extract
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco/optimizer: use opsel for VOP12C
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: don't label mul with opsel as abs/neg
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: swap opsel when swapping VOP2/C operands
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: support neg(mul)/abs(mul) optimization in more...
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: support v_cvt_f32_f16 with opsel in combine_mad_mix
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: update match_op3_for_vop3 for VOP12C opsel
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: handle opsel in combine_constant_comparison_ordering
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: handle opsel in combine_ordering_test
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco: handle opsel in combine_comparison_ordering
blob
|
commitdiff
|
diff to current
2023-03-30
Georg Lehmann
aco/optimizer: preserve opsel when fusing fma
blob
|
commitdiff
|
diff to current
2023-03-28
Georg Lehmann
aco: clean up to_mad_mix
blob
|
commitdiff
|
diff to current
2023-03-27
Georg Lehmann
aco: don't check usesModifiers for pseudo instructions
blob
|
commitdiff
|
diff to current
2023-03-17
Timur Kristóf
aco: Don't add soffset to swizzled MUBUF base.
blob
|
commitdiff
|
diff to current
2023-03-09
Georg Lehmann
aco: use bitfield_array for temporary neg/abs/opsel
blob
|
commitdiff
|
diff to current
2023-03-09
Georg Lehmann
aco: use array indexing for opsel/opsel_lo/opsel_hi
blob
|
commitdiff
|
diff to current
2023-03-09
Georg Lehmann
aco: use integer access for neg_lo/neg_hi
blob
|
commitdiff
|
diff to current
2023-03-09
Georg Lehmann
aco: copy abs/neg with assignment
blob
|
commitdiff
|
diff to current
2023-03-08
Georg Lehmann
aco: don't reallocate fma{mk,ak,_mix} instruction
blob
|
commitdiff
|
diff to current
2023-03-08
Georg Lehmann
aco/optimizer: don't reallocate instruction when conver...
blob
|
commitdiff
|
diff to current
2023-03-07
Georg Lehmann
aco: use bitfield array helpers for valu modifiers
blob
|
commitdiff
|
diff to current
2023-03-07
Georg Lehmann
aco: remove VOP[123C]P? structs
blob
|
commitdiff
|
diff to current
2023-03-07
Georg Lehmann
aco/optimizer: simplify using VALU instruction
blob
|
commitdiff
|
diff to current
2023-03-02
Georg Lehmann
aco: use v_fma_mix_f32 for v_fma_f32 with 2 fp16 repres...
blob
|
commitdiff
|
diff to current
2023-03-02
Georg Lehmann
aco: mark mad definition as precise if the mul/add...
blob
|
commitdiff
|
diff to current
2023-02-21
Rhys Perry
aco: don't apply modifiers through DPP to unsupported...
blob
|
commitdiff
|
diff to current
2023-02-21
Georg Lehmann
aco: combine a ^ ~b and ~(a ^ b) to v_xnor_b32
blob
|
commitdiff
|
diff to current
2023-02-18
Timur Kristóf
aco/optimizer: Change v_cmp with subgroup invocation...
blob
|
commitdiff
|
diff to current
2023-02-16
Timur Kristóf
aco: Remove MTBUF zero operand.
blob
|
commitdiff
|
diff to current
2023-02-08
Georg Lehmann
aco: add mov/cndmask opcodes to does_fp_op_flush_denorms
blob
|
commitdiff
|
diff to current
2023-01-23
Georg Lehmann
Revert "aco: Combine v_cvt_u32_f32 with insert to v_cvt...
blob
|
commitdiff
|
diff to current
2023-01-16
Timur Kristóf
aco/optimizer: Optimize p_extract + v_mul_u32_u24 to...
blob
|
commitdiff
|
diff to current
2023-01-16
Timur Kristóf
aco/optimizer: Add missing v_lshlrev condition to can_a...
blob
|
commitdiff
|
diff to current
2023-01-10
Rhys Perry
aco: disallow SGPRS/constants with interpolation instru...
blob
|
commitdiff
|
diff to current
2022-12-14
Timur Kristóf
aco: Emulate Wave64 bpermute on GFX11.
blob
|
commitdiff
|
diff to current
2022-12-14
Timur Kristóf
aco: Split opcodes for GFX6 and GFX10 emulated bpermute.
blob
|
commitdiff
|
diff to current
2022-12-14
Timur Kristóf
aco: Don't accept constants on p_bpermute.
blob
|
commitdiff
|
diff to current
2022-12-06
Rhys Perry
aco: more carefully apply constant offsets into scratch...
blob
|
commitdiff
|
diff to current
2022-12-01
Rhys Perry
aco: use v_minmax/v_maxmin opcodes
blob
|
commitdiff
|
diff to current
2022-12-01
Rhys Perry
aco: change order in combine_minmax()
blob
|
commitdiff
|
diff to current
2022-11-28
Georg Lehmann
aco: Combine constant bit test to s_bitcmp.
blob
|
commitdiff
|
diff to current
2022-11-28
Georg Lehmann
aco: Combine bit test to s_bitcmp.
blob
|
commitdiff
|
diff to current
2022-11-28
Georg Lehmann
aco: Combine s_abs and s_sub/s_add to s_absdiff.
blob
|
commitdiff
|
diff to current
2022-11-28
Georg Lehmann
aco: Ignore instructions with exec operands in follow_o...
blob
|
commitdiff
|
diff to current
2022-11-28
Georg Lehmann
aco/optimizer: Cleanup ctx.uses handling for patterns...
blob
|
commitdiff
|
diff to current
2022-11-24
Yonggang Luo
aco: Fixes -Werror,-Wbitwise-instead-of-logical for...
blob
|
commitdiff
|
diff to current
2022-11-16
Samuel Pitoiset
aco: add p_dual_src_export_gfx11 for dual source blendi...
blob
|
commitdiff
|
diff to current
2022-11-16
Georg Lehmann
aco: Combine v_cvt_u32_f32 with insert to v_cvt_pk_u8_f32.
blob
|
commitdiff
|
diff to current
2022-11-01
Rhys Perry
aco/gfx11: fix FS input loads in quad-divergent control...
blob
|
commitdiff
|
diff to current
2022-10-31
Georg Lehmann
aco: Use opsel for the third operand.
blob
|
commitdiff
|
diff to current
2022-10-24
Georg Lehmann
aco: Implement signed idot instructions on GFX11.
blob
|
commitdiff
|
diff to current
2022-10-24
Georg Lehmann
aco: Don't use opsel for p_insert.
blob
|
commitdiff
|
diff to current
2022-10-20
Timur Kristóf
aco: Optimize MUBUF 0 offset when idxen is also being...
blob
|
commitdiff
|
diff to current
2022-09-30
Rhys Perry
aco/gfx11: fix s_waitcnt printing
blob
|
commitdiff
|
diff to current
2022-09-22
Daniel Schürmann
aco/optimizer: change inverse_comparison in-place
blob
|
commitdiff
|
diff to current
2022-09-22
Timur Kristóf
aco: Change inverse-comparison optimization to work...
blob
|
commitdiff
|
diff to current
2022-09-22
Daniel Schürmann
aco/optimizer: optimize s_and(exec, s_and(x, y)) more...
blob
|
commitdiff
|
diff to current
2022-09-22
Daniel Schürmann
aco/optimizer: do can_eliminate_and_exec() optimization...
blob
|
commitdiff
|
diff to current
2022-09-21
Georg Lehmann
aco: Unswizzle v_pk_fma_f16 literals to produce more...
blob
|
commitdiff
|
diff to current
2022-09-21
Daniel Schürmann
aco/optimizer: disallow can_eliminate_and_exec() with...
blob
|
commitdiff
|
diff to current
2022-09-21
Georg Lehmann
aco: Use v_fmaak/v_fmamk if two operands are the same...
blob
|
commitdiff
|
diff to current
2022-08-30
Daniel Schürmann
aco: use std::vector::reserve() more often
blob
|
commitdiff
|
diff to current
2022-08-16
Rhys Perry
aco: rename is_cmp to is_fp_cmp
blob
|
commitdiff
|
diff to current
2022-07-11
Daniel Schürmann
aco: fix packed 16bit fneg/fsat optimization
blob
|
commitdiff
|
diff to current
2022-07-08
Rhys Perry
aco: handle subtractions in parse_base_offset
blob
|
commitdiff
|
diff to current
2022-07-08
Rhys Perry
aco: combine additions and constants into scratch load...
blob
|
commitdiff
|
diff to current
2022-07-08
Georg Lehmann
aco: Fix swapping sources in SOPC -> SOPK optimization.
blob
|
commitdiff
|
diff to current
2022-07-06
Georg Lehmann
aco/optimizer: Optimize SOPC with literal to SOPK.
blob
|
commitdiff
|
diff to current
2022-07-06
Georg Lehmann
aco/optimizer: Convert s_add_u32 with literals to s_add...
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: don't use 32-bit fp inline constants for fp16...
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: try sign-extending or shifting constants in propag...
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: fix redirect combine in propagate_constants_vop3p...
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: don't use opsel to fold constants into dot accumul...
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: fix propagate_constants_vop3p with integer vop3p...
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: ignore precise flag when optimizing integer clamps
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: include _e64 variants of 16-bit min/max in minmax...
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: don't accept med3 opcodes in get_minmax_info()
blob
|
commitdiff
|
diff to current
2022-07-05
Rhys Perry
aco: add and use constantValue16()
blob
|
commitdiff
|
diff to current
2022-06-06
Rhys Perry
aco: consider fma with multiplication by power-of-two...
blob
|
commitdiff
|
diff to current
2022-05-13
Marek Olšák
amd: change chip_class naming to "enum amd_gfx_level...
blob
|
commitdiff
|
diff to current
next