X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fglsl%2Fnir%2Fnir_opt_algebraic.py;h=6181f5df7a486b9fb7996c21e0d96b78c1753ce6;hb=97e6c1b9;hp=9c62b2836340e37f2fb9020c98fe71b746ef4fc2;hpb=15063d2ad0945004ef387511d0c805bbe734bc5f;p=mesa.git

diff --git a/src/glsl/nir/nir_opt_algebraic.py b/src/glsl/nir/nir_opt_algebraic.py
index 9c62b283634..6181f5df7a4 100644
--- a/src/glsl/nir/nir_opt_algebraic.py
+++ b/src/glsl/nir/nir_opt_algebraic.py
@@ -36,9 +36,15 @@ d = 'd'
 # and <replace> is either an expression or a value.  An expression is
 # defined as a tuple of the form (<op>, <src0>, <src1>, <src2>, <src3>)
 # where each source is either an expression or a value.  A value can be
-# either a numeric constant or a string representing a variable name.  For
-# constants, you have to be careful to make sure that it is the right type
-# because python is unaware of the source and destination types of the
+# either a numeric constant or a string representing a variable name.
+#
+# Variable names are specified as "[#]name[@type]" where "#" inicates that
+# the given variable will only match constants and the type indicates that
+# the given variable will only match values from ALU instructions with the
+# given output type.
+#
+# For constants, you have to be careful to make sure that it is the right
+# type because python is unaware of the source and destination types of the
 # opcodes.
 
 optimizations = [
@@ -50,6 +56,10 @@ optimizations = [
    (('iabs', ('ineg', a)), ('iabs', a)),
    (('fadd', a, 0.0), a),
    (('iadd', a, 0), a),
+   (('fadd', ('fmul', a, b), ('fmul', a, c)), ('fmul', a, ('fadd', b, c))),
+   (('iadd', ('imul', a, b), ('imul', a, c)), ('imul', a, ('iadd', b, c))),
+   (('fadd', ('fneg', a), a), 0.0),
+   (('iadd', ('ineg', a), a), 0),
    (('fmul', a, 0.0), 0.0),
    (('imul', a, 0), 0),
    (('fmul', a, 1.0), a),
@@ -65,18 +75,42 @@ optimizations = [
    (('flrp', a, b, 1.0), b),
    (('flrp', a, a, b), a),
    (('flrp', 0.0, a, b), ('fmul', a, b)),
-   (('fadd', ('fmul', a, b), c), ('ffma', a, b, c)),
+   (('flrp', a, b, c), ('fadd', ('fmul', c, ('fsub', b, a)), a), 'options->lower_flrp'),
+   (('fadd', ('fmul', a, ('fadd', 1.0, ('fneg', c))), ('fmul', b, c)), ('flrp', a, b, c), '!options->lower_flrp'),
+   (('fadd', a, ('fmul', c, ('fadd', b, ('fneg', a)))), ('flrp', a, b, c), '!options->lower_flrp'),
+   (('ffma', a, b, c), ('fadd', ('fmul', a, b), c), 'options->lower_ffma'),
+   (('fadd', ('fmul', a, b), c), ('ffma', a, b, c), '!options->lower_ffma'),
    # Comparison simplifications
    (('inot', ('flt', a, b)), ('fge', a, b)),
    (('inot', ('fge', a, b)), ('flt', a, b)),
    (('inot', ('ilt', a, b)), ('ige', a, b)),
    (('inot', ('ige', a, b)), ('ilt', a, b)),
-   (('flt', ('fadd', a, b), 0.0), ('flt', a, ('fneg', b))),
-   (('fge', ('fadd', a, b), 0.0), ('fge', a, ('fneg', b))),
-   (('feq', ('fadd', a, b), 0.0), ('feq', a, ('fneg', b))),
-   (('fne', ('fadd', a, b), 0.0), ('fne', a, ('fneg', b))),
    (('fge', ('fneg', ('fabs', a)), 0.0), ('feq', a, 0.0)),
-   (('fmin', ('fmax', a, 1.0), 0.0), ('fsat', a)),
+   (('bcsel', ('flt', a, b), a, b), ('fmin', a, b)),
+   (('bcsel', ('flt', a, b), b, a), ('fmax', a, b)),
+   (('bcsel', ('inot', 'a@bool'), b, c), ('bcsel', a, c, b)),
+   (('bcsel', a, ('bcsel', a, b, c), d), ('bcsel', a, b, d)),
+   (('fmin', ('fmax', a, 0.0), 1.0), ('fsat', a), '!options->lower_fsat'),
+   (('fsat', a), ('fmin', ('fmax', a, 0.0), 1.0), 'options->lower_fsat'),
+   (('fsat', ('fsat', a)), ('fsat', a)),
+   (('fmin', ('fmax', ('fmin', ('fmax', a, 0.0), 1.0), 0.0), 1.0), ('fmin', ('fmax', a, 0.0), 1.0)),
+   (('ior', ('flt', a, b), ('flt', a, c)), ('flt', a, ('fmax', b, c))),
+   (('ior', ('fge', a, b), ('fge', a, c)), ('fge', a, ('fmin', b, c))),
+   # Emulating booleans
+   (('fmul', ('b2f', a), ('b2f', b)), ('b2f', ('iand', a, b))),
+   (('fsat', ('fadd', ('b2f', a), ('b2f', b))), ('b2f', ('ior', a, b))),
+   (('iand', 'a@bool', 1.0), ('b2f', 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.
+   # Comparison with the same args.  Note that these are not done for
+   # the float versions because NaN always returns false on float
+   # inequalities.
+   (('ilt', a, a), False),
+   (('ige', a, a), True),
+   (('ieq', a, a), True),
+   (('ine', a, a), False),
+   (('ult', a, a), False),
+   (('uge', a, a), True),
    # Logical and bit operations
    (('fand', a, 0.0), 0.0),
    (('iand', a, a), a),
@@ -101,19 +135,97 @@ optimizations = [
    (('fexp',  ('flog',  a)), a), # e^ln(a)  = a
    (('flog2', ('fexp2', a)), a), # lg2(2^a) = a
    (('flog',  ('fexp',  a)), a), # ln(e^a)  = a
-   (('fexp2', ('fmul', ('flog2', a), b)), ('fpow', a, b)), # 2^(lg2(a)*b) = a^b
-   (('fexp',  ('fmul', ('flog', a), b)),  ('fpow', a, b)), # e^(ln(a)*b) = a^b
+   (('fpow', a, b), ('fexp2', ('fmul', ('flog2', a), b)), 'options->lower_fpow'), # a^b = 2^(lg2(a)*b)
+   (('fexp2', ('fmul', ('flog2', a), b)), ('fpow', a, b), '!options->lower_fpow'), # 2^(lg2(a)*b) = a^b
+   (('fexp',  ('fmul', ('flog', a), b)),  ('fpow', a, b), '!options->lower_fpow'), # e^(ln(a)*b) = a^b
    (('fpow', a, 1.0), a),
    (('fpow', a, 2.0), ('fmul', a, a)),
    (('fpow', 2.0, a), ('fexp2', a)),
+   (('fsqrt', ('fexp2', a)), ('fexp2', ('fmul', 0.5, a))),
+   (('fsqrt', ('fexp', a)), ('fexp', ('fmul', 0.5, a))),
+   (('frcp', ('fexp2', a)), ('fexp2', ('fneg', a))),
+   (('frcp', ('fexp', a)), ('fexp', ('fneg', a))),
+   (('frsq', ('fexp2', a)), ('fexp2', ('fmul', -0.5, a))),
+   (('frsq', ('fexp', a)), ('fexp', ('fmul', -0.5, a))),
+   (('flog2', ('fsqrt', a)), ('fmul', 0.5, ('flog2', a))),
+   (('flog', ('fsqrt', a)), ('fmul', 0.5, ('flog', a))),
+   (('flog2', ('frcp', a)), ('fneg', ('flog2', a))),
+   (('flog', ('frcp', a)), ('fneg', ('flog', a))),
+   (('flog2', ('frsq', a)), ('fmul', -0.5, ('flog2', a))),
+   (('flog', ('frsq', a)), ('fmul', -0.5, ('flog', a))),
+   (('flog2', ('fpow', a, b)), ('fmul', b, ('flog2', a))),
+   (('flog', ('fpow', a, b)), ('fmul', b, ('flog', a))),
+   (('fadd', ('flog2', a), ('flog2', b)), ('flog2', ('fmul', a, b))),
+   (('fadd', ('flog', a), ('flog', b)), ('flog', ('fmul', a, b))),
+   (('fadd', ('flog2', a), ('fneg', ('flog2', b))), ('flog2', ('fdiv', a, b))),
+   (('fadd', ('flog', a), ('fneg', ('flog', b))), ('flog', ('fdiv', a, b))),
+   (('fmul', ('fexp2', a), ('fexp2', b)), ('fexp2', ('fadd', a, b))),
+   (('fmul', ('fexp', a), ('fexp', b)), ('fexp', ('fadd', a, b))),
    # Division and reciprocal
    (('fdiv', 1.0, a), ('frcp', a)),
    (('frcp', ('frcp', a)), a),
    (('frcp', ('fsqrt', a)), ('frsq', a)),
-   (('frcp', ('frsq', a)), ('fsqrt', a)),
+   (('fsqrt', a), ('frcp', ('frsq', a)), 'options->lower_fsqrt'),
+   (('frcp', ('frsq', a)), ('fsqrt', a), '!options->lower_fsqrt'),
+   # Boolean simplifications
+   (('ine', 'a@bool', 0), 'a'),
+   (('ieq', 'a@bool', 0), ('inot', 'a')),
+   (('bcsel', a, True, False), ('ine', a, 0)),
+   (('bcsel', a, False, True), ('ieq', a, 0)),
+   (('bcsel', True, b, c), b),
+   (('bcsel', False, b, c), 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)),
+
+   (('bcsel', a, b, b), b),
+   (('fcsel', a, b, b), b),
+
+   # Subtracts
+   (('fsub', a, ('fsub', 0.0, b)), ('fadd', a, b)),
+   (('isub', a, ('isub', 0, b)), ('iadd', a, b)),
+   (('fsub', a, b), ('fadd', a, ('fneg', b)), '!options->lower_negate'),
+   (('isub', a, b), ('iadd', a, ('ineg', b)), '!options->lower_negate'),
+   (('fneg', a), ('fsub', 0.0, a), 'options->lower_negate'),
+   (('ineg', a), ('isub', 0, a), 'options->lower_negate'),
+   (('fadd', a, ('fsub', 0.0, b)), ('fsub', a, b)),
+   (('iadd', a, ('isub', 0, b)), ('isub', a, b)),
+   (('fabs', ('fsub', 0.0, a)), ('fabs', a)),
+   (('iabs', ('isub', 0, a)), ('iabs', a)),
+]
 
-# This one may not be exact
+# Add optimizations to handle the case where the result of a ternary is
+# compared to a constant.  This way we can take things like
+#
+# (a ? 0 : 1) > 0
+#
+# and turn it into
+#
+# a ? (0 > 0) : (1 > 0)
+#
+# which constant folding will eat for lunch.  The resulting ternary will
+# further get cleaned up by the boolean reductions above and we will be
+# left with just the original variable "a".
+for op in ['flt', 'fge', 'feq', 'fne',
+           'ilt', 'ige', 'ieq', 'ine', 'ult', 'uge']:
+   optimizations += [
+      ((op, ('bcsel', 'a', '#b', '#c'), '#d'),
+       ('bcsel', 'a', (op, 'b', 'd'), (op, 'c', 'd'))),
+      ((op, '#d', ('bcsel', a, '#b', '#c')),
+       ('bcsel', 'a', (op, 'd', 'b'), (op, 'd', 'c'))),
+   ]
+
+# This section contains "late" optimizations that should be run after the
+# regular optimizations have finished.  Optimizations should go here if
+# they help code generation but do not necessarily produce code that is
+# more easily optimizable.
+late_optimizations = [
+   (('flt', ('fadd', a, b), 0.0), ('flt', a, ('fneg', b))),
+   (('fge', ('fadd', a, b), 0.0), ('fge', a, ('fneg', b))),
    (('feq', ('fadd', a, b), 0.0), ('feq', a, ('fneg', b))),
+   (('fne', ('fadd', a, b), 0.0), ('fne', a, ('fneg', b))),
 ]
 
 print nir_algebraic.AlgebraicPass("nir_opt_algebraic", optimizations).render()
+print nir_algebraic.AlgebraicPass("nir_opt_algebraic_late",
+                                  late_optimizations).render()