diff options
| -rw-r--r-- | src/compiler/nir/nir_opt_algebraic.py | 13 | ||||
| -rw-r--r-- | src/compiler/nir/nir_search_helpers.h | 16 |
2 files changed, 24 insertions, 5 deletions
diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py index 79058c06b97..1c7b3597c1f 100644 --- a/src/compiler/nir/nir_opt_algebraic.py +++ b/src/compiler/nir/nir_opt_algebraic.py @@ -152,11 +152,7 @@ optimizations = [ (('~fadd@32', ('fmul', a, ('fadd', 1.0, ('fneg', c ) )), ('fmul', b, c )), ('flrp', a, b, c), '!options->lower_flrp32'), (('~fadd@64', ('fmul', a, ('fadd', 1.0, ('fneg', c ) )), ('fmul', b, c )), ('flrp', a, b, c), '!options->lower_flrp64'), # These are the same as the previous three rules, but it depends on - # 1-fsat(x) <=> fsat(1-x): - # - # If x >= 0 and x <= 1: fsat(1 - x) == 1 - fsat(x) trivially - # If x < 0: 1 - fsat(x) => 1 - 0 => 1 and fsat(1 - x) => fsat(> 1) => 1 - # If x > 1: 1 - fsat(x) => 1 - 1 => 0 and fsat(1 - x) => fsat(< 0) => 0 + # 1-fsat(x) <=> fsat(1-x). See below. (('~fadd@32', ('fmul', a, ('fsat', ('fadd', 1.0, ('fneg', c )))), ('fmul', b, ('fsat', c))), ('flrp', a, b, ('fsat', c)), '!options->lower_flrp32'), (('~fadd@64', ('fmul', a, ('fsat', ('fadd', 1.0, ('fneg', c )))), ('fmul', b, ('fsat', c))), ('flrp', a, b, ('fsat', c)), '!options->lower_flrp64'), @@ -177,6 +173,11 @@ optimizations = [ (('fdot3', ('vec3', a, 0.0, 0.0), b), ('fmul', a, b)), (('fdot3', ('vec3', a, b, 0.0), c), ('fdot2', ('vec2', a, b), c)), + # If x >= 0 and x <= 1: fsat(1 - x) == 1 - fsat(x) trivially + # If x < 0: 1 - fsat(x) => 1 - 0 => 1 and fsat(1 - x) => fsat(> 1) => 1 + # If x > 1: 1 - fsat(x) => 1 - 1 => 0 and fsat(1 - x) => fsat(< 0) => 0 + (('~fadd', ('fneg(is_used_once)', ('fsat(is_used_once)', 'a(is_not_fmul)')), 1.0), ('fsat', ('fadd', 1.0, ('fneg', a)))), + # (a * #b + #c) << #d # ((a * #b) << #d) + (#c << #d) # (a * (#b << #d)) + (#c << #d) @@ -1179,6 +1180,8 @@ late_optimizations = [ (('ior', a, a), a), (('iand', a, a), a), + (('~fadd', ('fneg(is_used_once)', ('fsat(is_used_once)', 'a(is_not_fmul)')), 1.0), ('fsat', ('fadd', 1.0, ('fneg', a)))), + (('fdot2', a, b), ('fdot_replicated2', a, b), 'options->fdot_replicates'), (('fdot3', a, b), ('fdot_replicated3', a, b), 'options->fdot_replicates'), (('fdot4', a, b), ('fdot_replicated4', a, b), 'options->fdot_replicates'), diff --git a/src/compiler/nir/nir_search_helpers.h b/src/compiler/nir/nir_search_helpers.h index 9e03c610ece..631c65a8642 100644 --- a/src/compiler/nir/nir_search_helpers.h +++ b/src/compiler/nir/nir_search_helpers.h @@ -174,6 +174,22 @@ is_not_const(nir_alu_instr *instr, unsigned src, UNUSED unsigned num_components, } static inline bool +is_not_fmul(nir_alu_instr *instr, unsigned src, + UNUSED unsigned num_components, UNUSED const uint8_t *swizzle) +{ + nir_alu_instr *src_alu = + nir_src_as_alu_instr(instr->src[src].src); + + if (src_alu == NULL) + return true; + + if (src_alu->op == nir_op_fneg) + return is_not_fmul(src_alu, 0, 0, NULL); + + return src_alu->op != nir_op_fmul; +} + +static inline bool is_used_once(nir_alu_instr *instr) { bool zero_if_use = list_empty(&instr->dest.dest.ssa.if_uses); |