[Mesa-dev] [PATCH v2 08/11] nir/opt_algebraic: Add 32-bit specifiers to a bunch of booleans
Jason Ekstrand
jason at jlekstrand.net
Thu Nov 29 23:37:20 UTC 2018
---
src/compiler/nir/nir_opt_algebraic.py | 112 +++++++++++++-------------
1 file changed, 56 insertions(+), 56 deletions(-)
diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py
index c482bde8b3b..aa1a7a94e6e 100644
--- a/src/compiler/nir/nir_opt_algebraic.py
+++ b/src/compiler/nir/nir_opt_algebraic.py
@@ -119,17 +119,17 @@ optimizations = [
(('~flrp', a, b, 1.0), b),
(('~flrp', a, a, b), a),
(('~flrp', 0.0, a, b), ('fmul', a, b)),
- (('~flrp', a, b, ('b2f', c)), ('bcsel', c, b, a), 'options->lower_flrp32'),
+ (('~flrp', a, b, ('b2f', 'c at 32')), ('bcsel', c, b, a), 'options->lower_flrp32'),
(('~flrp', a, 0.0, c), ('fadd', ('fmul', ('fneg', a), c), a)),
(('flrp at 32', a, b, c), ('fadd', ('fmul', c, ('fsub', b, a)), a), 'options->lower_flrp32'),
(('flrp at 64', a, b, c), ('fadd', ('fmul', c, ('fsub', b, a)), a), 'options->lower_flrp64'),
(('ffloor', a), ('fsub', a, ('ffract', a)), 'options->lower_ffloor'),
(('ffract', a), ('fsub', a, ('ffloor', a)), 'options->lower_ffract'),
(('fceil', a), ('fneg', ('ffloor', ('fneg', a))), 'options->lower_fceil'),
- (('~fadd', ('fmul', a, ('fadd', 1.0, ('fneg', ('b2f', c)))), ('fmul', b, ('b2f', c))), ('bcsel', c, b, a), 'options->lower_flrp32'),
+ (('~fadd', ('fmul', a, ('fadd', 1.0, ('fneg', ('b2f', 'c at 32')))), ('fmul', b, ('b2f', c))), ('bcsel', c, b, a), 'options->lower_flrp32'),
(('~fadd at 32', ('fmul', a, ('fadd', 1.0, ('fneg', c ))), ('fmul', b, c )), ('flrp', a, b, c), '!options->lower_flrp32'),
(('~fadd at 64', ('fmul', a, ('fadd', 1.0, ('fneg', c ))), ('fmul', b, c )), ('flrp', a, b, c), '!options->lower_flrp64'),
- (('~fadd', a, ('fmul', ('b2f', c), ('fadd', b, ('fneg', a)))), ('bcsel', c, b, a), 'options->lower_flrp32'),
+ (('~fadd', a, ('fmul', ('b2f', 'c at 32'), ('fadd', b, ('fneg', a)))), ('bcsel', c, b, a), 'options->lower_flrp32'),
(('~fadd at 32', a, ('fmul', c , ('fadd', b, ('fneg', a)))), ('flrp', a, b, c), '!options->lower_flrp32'),
(('~fadd at 64', a, ('fmul', c , ('fadd', b, ('fneg', a)))), ('flrp', a, b, c), '!options->lower_flrp64'),
(('ffma', a, b, c), ('fadd', ('fmul', a, b), c), 'options->lower_ffma'),
@@ -169,50 +169,50 @@ optimizations = [
# b2f(a) <= 0.0
# b2f(a) == 0.0 because b2f(a) can only be 0 or 1
# inot(a)
- (('fge', 0.0, ('b2f', a)), ('inot', a)),
-
- (('fge', ('fneg', ('b2f', a)), 0.0), ('inot', a)),
-
- (('fne', ('fadd', ('b2f', a), ('b2f', b)), 0.0), ('ior', a, b)),
- (('fne', ('fmax', ('b2f', a), ('b2f', b)), 0.0), ('ior', a, b)),
- (('fne', ('bcsel', a, 1.0, ('b2f', b)) , 0.0), ('ior', a, b)),
- (('fne', ('b2f', a), ('fneg', ('b2f', b))), ('ior', a, b)),
- (('fne', ('fmul', ('b2f', a), ('b2f', b)), 0.0), ('iand', a, b)),
- (('fne', ('fmin', ('b2f', a), ('b2f', b)), 0.0), ('iand', a, b)),
- (('fne', ('bcsel', a, ('b2f', b), 0.0) , 0.0), ('iand', a, b)),
- (('fne', ('fadd', ('b2f', a), ('fneg', ('b2f', b))), 0.0), ('ixor', a, b)),
- (('fne', ('b2f', a) , ('b2f', b) ), ('ixor', a, b)),
- (('fne', ('fneg', ('b2f', a)), ('fneg', ('b2f', b))), ('ixor', a, b)),
- (('feq', ('fadd', ('b2f', a), ('b2f', b)), 0.0), ('inot', ('ior', a, b))),
- (('feq', ('fmax', ('b2f', a), ('b2f', b)), 0.0), ('inot', ('ior', a, b))),
- (('feq', ('bcsel', a, 1.0, ('b2f', b)) , 0.0), ('inot', ('ior', a, b))),
- (('feq', ('b2f', a), ('fneg', ('b2f', b))), ('inot', ('ior', a, b))),
- (('feq', ('fmul', ('b2f', a), ('b2f', b)), 0.0), ('inot', ('iand', a, b))),
- (('feq', ('fmin', ('b2f', a), ('b2f', b)), 0.0), ('inot', ('iand', a, b))),
- (('feq', ('bcsel', a, ('b2f', b), 0.0) , 0.0), ('inot', ('iand', a, b))),
- (('feq', ('fadd', ('b2f', a), ('fneg', ('b2f', b))), 0.0), ('ieq', a, b)),
- (('feq', ('b2f', a) , ('b2f', b) ), ('ieq', a, b)),
- (('feq', ('fneg', ('b2f', a)), ('fneg', ('b2f', b))), ('ieq', a, b)),
+ (('fge', 0.0, ('b2f', 'a at 32')), ('inot', a)),
+
+ (('fge', ('fneg', ('b2f', 'a at 32')), 0.0), ('inot', a)),
+
+ (('fne', ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('ior', a, b)),
+ (('fne', ('fmax', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('ior', a, b)),
+ (('fne', ('bcsel', a, 1.0, ('b2f', 'b at 32')) , 0.0), ('ior', a, b)),
+ (('fne', ('b2f', 'a at 32'), ('fneg', ('b2f', 'b at 32'))), ('ior', a, b)),
+ (('fne', ('fmul', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('iand', a, b)),
+ (('fne', ('fmin', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('iand', a, b)),
+ (('fne', ('bcsel', a, ('b2f', 'b at 32'), 0.0) , 0.0), ('iand', a, b)),
+ (('fne', ('fadd', ('b2f', 'a at 32'), ('fneg', ('b2f', 'b at 32'))), 0.0), ('ixor', a, b)),
+ (('fne', ('b2f', 'a at 32') , ('b2f', 'b at 32') ), ('ixor', a, b)),
+ (('fne', ('fneg', ('b2f', 'a at 32')), ('fneg', ('b2f', 'b at 32'))), ('ixor', a, b)),
+ (('feq', ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('inot', ('ior', a, b))),
+ (('feq', ('fmax', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('inot', ('ior', a, b))),
+ (('feq', ('bcsel', a, 1.0, ('b2f', 'b at 32')) , 0.0), ('inot', ('ior', a, b))),
+ (('feq', ('b2f', 'a at 32'), ('fneg', ('b2f', 'b at 32'))), ('inot', ('ior', a, b))),
+ (('feq', ('fmul', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('inot', ('iand', a, b))),
+ (('feq', ('fmin', ('b2f', 'a at 32'), ('b2f', 'b at 32')), 0.0), ('inot', ('iand', a, b))),
+ (('feq', ('bcsel', a, ('b2f', 'b at 32'), 0.0) , 0.0), ('inot', ('iand', a, b))),
+ (('feq', ('fadd', ('b2f', 'a at 32'), ('fneg', ('b2f', 'b at 32'))), 0.0), ('ieq', a, b)),
+ (('feq', ('b2f', 'a at 32') , ('b2f', 'b at 32') ), ('ieq', a, b)),
+ (('feq', ('fneg', ('b2f', 'a at 32')), ('fneg', ('b2f', 'b at 32'))), ('ieq', a, b)),
# -(b2f(a) + b2f(b)) < 0
# 0 < b2f(a) + b2f(b)
# 0 != b2f(a) + b2f(b) b2f must be 0 or 1, so the sum is non-negative
# a || b
- (('flt', ('fneg', ('fadd', ('b2f', a), ('b2f', b))), 0.0), ('ior', a, b)),
- (('flt', 0.0, ('fadd', ('b2f', a), ('b2f', b))), ('ior', a, b)),
+ (('flt', ('fneg', ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32'))), 0.0), ('ior', a, b)),
+ (('flt', 0.0, ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32'))), ('ior', a, b)),
# -(b2f(a) + b2f(b)) >= 0
# 0 >= b2f(a) + b2f(b)
# 0 == b2f(a) + b2f(b) b2f must be 0 or 1, so the sum is non-negative
# !(a || b)
- (('fge', ('fneg', ('fadd', ('b2f', a), ('b2f', b))), 0.0), ('inot', ('ior', a, b))),
- (('fge', 0.0, ('fadd', ('b2f', a), ('b2f', b))), ('inot', ('ior', a, b))),
+ (('fge', ('fneg', ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32'))), 0.0), ('inot', ('ior', a, b))),
+ (('fge', 0.0, ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32'))), ('inot', ('ior', a, b))),
# Some optimizations (below) convert things like (a < b || c < b) into
# (min(a, c) < b). However, this interfers with the previous optimizations
# that try to remove comparisons with negated sums of b2f. This just
# breaks that apart.
- (('flt', ('fmin', c, ('fneg', ('fadd', ('b2f', a), ('b2f', b)))), 0.0),
+ (('flt', ('fmin', c, ('fneg', ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32')))), 0.0),
('ior', ('flt', c, 0.0), ('ior', a, b))),
(('~flt', ('fadd', a, b), a), ('flt', b, 0.0)),
@@ -234,13 +234,13 @@ optimizations = [
# The fge in the second replacement is not a typo. I leave the proof that
# "fmin(-b2f(a), b) >= 0 <=> fmin(-b2f(a), b) == 0" as an exercise for the
# reader.
- (('fge', ('fmin', ('fneg', ('b2f', a)), b), 0.0), ('iand', ('inot', a), ('fge', b, 0.0))),
- (('feq', ('fmin', ('fneg', ('b2f', a)), b), 0.0), ('iand', ('inot', a), ('fge', b, 0.0))),
+ (('fge', ('fmin', ('fneg', ('b2f', 'a at 32')), 'b at 32'), 0.0), ('iand', ('inot', a), ('fge', b, 0.0))),
+ (('feq', ('fmin', ('fneg', ('b2f', 'a at 32')), 'b at 32'), 0.0), ('iand', ('inot', a), ('fge', b, 0.0))),
- (('feq', ('b2f', a), 0.0), ('inot', a)),
- (('fne', ('b2f', a), 0.0), a),
- (('ieq', ('b2i', a), 0), ('inot', a)),
- (('ine', ('b2i', a), 0), a),
+ (('feq', ('b2f', 'a at 32'), 0.0), ('inot', a)),
+ (('fne', ('b2f', 'a at 32'), 0.0), a),
+ (('ieq', ('b2i', 'a at 32'), 0), ('inot', a)),
+ (('ine', ('b2i', 'a at 32'), 0), a),
(('fne', ('u2f', a), 0.0), ('ine', a, 0)),
(('feq', ('u2f', a), 0.0), ('ieq', a, 0)),
@@ -274,10 +274,10 @@ optimizations = [
# 0.0 >= fabs(a)
(('fge', ('fneg', ('fabs', a)), 0.0), ('feq', a, 0.0)),
- (('fmax', ('b2f(is_used_once)', a), ('b2f', b)), ('b2f', ('ior', a, b))),
- (('fmax', ('fneg(is_used_once)', ('b2f(is_used_once)', a)), ('fneg', ('b2f', b))), ('fneg', ('b2f', ('ior', a, b)))),
- (('fmin', ('b2f(is_used_once)', a), ('b2f', b)), ('b2f', ('iand', a, b))),
- (('fmin', ('fneg(is_used_once)', ('b2f(is_used_once)', a)), ('fneg', ('b2f', b))), ('fneg', ('b2f', ('iand', a, b)))),
+ (('fmax', ('b2f(is_used_once)', 'a at 32'), ('b2f', 'b at 32')), ('b2f', ('ior', a, b))),
+ (('fmax', ('fneg(is_used_once)', ('b2f(is_used_once)', 'a at 32')), ('fneg', ('b2f', 'b at 32'))), ('fneg', ('b2f', ('ior', a, b)))),
+ (('fmin', ('b2f(is_used_once)', 'a at 32'), ('b2f', 'b at 32')), ('b2f', ('iand', a, b))),
+ (('fmin', ('fneg(is_used_once)', ('b2f(is_used_once)', 'a at 32')), ('fneg', ('b2f', 'b at 32'))), ('fneg', ('b2f', ('iand', a, b)))),
# fmin(b2f(a), b)
# bcsel(a, fmin(b2f(a), b), fmin(b2f(a), b))
@@ -286,7 +286,7 @@ optimizations = [
#
# Since b is a constant, constant folding will eliminate the fmin and the
# fmax. If b is > 1.0, the bcsel will be replaced with a b2f.
- (('fmin', ('b2f', a), '#b'), ('bcsel', a, ('fmin', b, 1.0), ('fmin', b, 0.0))),
+ (('fmin', ('b2f', 'a at 32'), '#b'), ('bcsel', a, ('fmin', b, 1.0), ('fmin', b, 0.0))),
(('flt', ('fadd(is_used_once)', a, ('fneg', b)), 0.0), ('flt', a, b)),
@@ -435,14 +435,14 @@ optimizations = [
(('fne', ('fneg', a), a), ('fne', a, 0.0)),
(('feq', ('fneg', a), a), ('feq', a, 0.0)),
# Emulating booleans
- (('imul', ('b2i', a), ('b2i', b)), ('b2i', ('iand', a, b))),
- (('fmul', ('b2f', a), ('b2f', b)), ('b2f', ('iand', a, b))),
- (('fsat', ('fadd', ('b2f', a), ('b2f', b))), ('b2f', ('ior', a, b))),
+ (('imul', ('b2i', 'a at 32'), ('b2i', 'b at 32')), ('b2i', ('iand', a, b))),
+ (('fmul', ('b2f', 'a at 32'), ('b2f', 'b at 32')), ('b2f', ('iand', a, b))),
+ (('fsat', ('fadd', ('b2f', 'a at 32'), ('b2f', 'b at 32'))), ('b2f', ('ior', a, b))),
(('iand', 'a at bool', 1.0), ('b2f', a), '!options->lower_b2f'),
# True/False are ~0 and 0 in NIR. b2i of True is 1, and -1 is ~0 (True).
- (('ineg', ('b2i at 32', a)), a),
- (('flt', ('fneg', ('b2f', a)), 0), a), # Generated by TGSI KILL_IF.
- (('flt', ('fsub', 0.0, ('b2f', a)), 0), a), # Generated by TGSI KILL_IF.
+ (('ineg', ('b2i at 32', 'a at 32')), a),
+ (('flt', ('fneg', ('b2f', 'a at 32')), 0), a), # Generated by TGSI KILL_IF.
+ (('flt', ('fsub', 0.0, ('b2f', 'a at 32')), 0), a), # Generated by TGSI KILL_IF.
# Comparison with the same args. Note that these are not done for
# the float versions because NaN always returns false on float
# inequalities.
@@ -523,7 +523,7 @@ optimizations = [
(('bcsel at 32', a, -0.0, -1.0), ('fneg', ('b2f', ('inot', a)))),
(('bcsel', True, b, c), b),
(('bcsel', False, b, c), c),
- (('bcsel', a, ('b2f(is_used_once)', b), ('b2f', c)), ('b2f', ('bcsel', a, b, c))),
+ (('bcsel', a, ('b2f(is_used_once)', 'b at 32'), ('b2f', 'c at 32')), ('b2f', ('bcsel', a, b, c))),
# The result of this should be hit by constant propagation and, in the
# next round of opt_algebraic, get picked up by one of the above two.
(('bcsel', '#a', b, c), ('bcsel', ('ine', 'a', 0), b, c)),
@@ -532,7 +532,7 @@ optimizations = [
(('fcsel', a, b, b), b),
# Conversions
- (('i2b', ('b2i', a)), a),
+ (('i2b', ('b2i', 'a at 32')), a),
(('i2b', 'a at bool'), a),
(('f2i', ('ftrunc', a)), ('f2i', a)),
(('f2u', ('ftrunc', a)), ('f2u', a)),
@@ -754,8 +754,8 @@ for left, right in itertools.combinations_with_replacement(invert.keys(), 2):
('ior', (invert[left], a, b), (invert[right], c, d))))
# Optimize x2yN(b2x(x)) -> b2y
-optimizations.append((('f2b', ('b2f', a)), a))
-optimizations.append((('i2b', ('b2i', a)), a))
+optimizations.append((('f2b', ('b2f', 'a at 32')), a))
+optimizations.append((('i2b', ('b2i', 'a at 32')), a))
for x, y in itertools.product(['f', 'u', 'i'], ['f', 'u', 'i']):
if x != 'f' and y != 'f' and x != y:
continue
@@ -908,15 +908,15 @@ late_optimizations = [
(('fdot4', a, b), ('fdot_replicated4', a, b), 'options->fdot_replicates'),
(('fdph', a, b), ('fdph_replicated', a, b), 'options->fdot_replicates'),
- (('b2f(is_used_more_than_once)', ('inot', a)), ('bcsel', a, 0.0, 1.0)),
- (('fneg(is_used_more_than_once)', ('b2f', ('inot', a))), ('bcsel', a, -0.0, -1.0)),
+ (('b2f(is_used_more_than_once)', ('inot', 'a at 32')), ('bcsel', a, 0.0, 1.0)),
+ (('fneg(is_used_more_than_once)', ('b2f', ('inot', 'a at 32'))), ('bcsel', a, -0.0, -1.0)),
# we do these late so that we don't get in the way of creating ffmas
(('fmin', ('fadd(is_used_once)', '#c', a), ('fadd(is_used_once)', '#c', b)), ('fadd', c, ('fmin', a, b))),
(('fmax', ('fadd(is_used_once)', '#c', a), ('fadd(is_used_once)', '#c', b)), ('fadd', c, ('fmax', a, b))),
# Lowered for backends without a dedicated b2f instruction
- (('b2f at 32', a), ('iand', a, 1.0), 'options->lower_b2f'),
+ (('b2f at 32', 'a at 32'), ('iand', a, 1.0), 'options->lower_b2f'),
]
print(nir_algebraic.AlgebraicPass("nir_opt_algebraic", optimizations).render())
--
2.19.1
More information about the mesa-dev
mailing list