nir/i965: add before ffma algebraic opts

This shuffles constants down in the reverse of what the previous
patch does and applies some simpilifications that may be made
possible from doing so.

Shader-db results BDW:

total instructions in shared programs: 12980814 -> 12977822 (-0.02%)
instructions in affected programs: 281889 -> 278897 (-1.06%)
helped: 1231
HURT: 128

total cycles in shared programs: 246562852 -> 246567288 (0.00%)
cycles in affected programs: 11271524 -> 11275960 (0.04%)
helped: 1630
HURT: 1378

V2: mark float opts as inexact

Reviewed-by: Elie Tournier <elie.tournier@collabora.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

diff --git a/src/compiler/nir/nir.h b/src/compiler/nir/nir.h
index ce5b434d56a2226d0d763743cabaaf435ed65725..dbbf26859fb8261c47ba4e10b893b71fbff999e2 100644 (file)
--- a/src/compiler/nir/nir.h
+++ b/src/compiler/nir/nir.h
@@ -2598,6 +2598,7 @@ bool nir_lower_phis_to_regs_block(nir_block *block);
 bool nir_lower_ssa_defs_to_regs_block(nir_block *block);
 
 bool nir_opt_algebraic(nir_shader *shader);
+bool nir_opt_algebraic_before_ffma(nir_shader *shader);
 bool nir_opt_algebraic_late(nir_shader *shader);
 bool nir_opt_constant_folding(nir_shader *shader);
 

diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py
index 974345a17e56af074df22c6e1dd8cd5fe60e042d..fe6e33d31397929ed169c242b510b45d146e8c79 100644 (file)
--- a/src/compiler/nir/nir_opt_algebraic.py
+++ b/src/compiler/nir/nir_opt_algebraic.py
@@ -530,6 +530,27 @@ for op in ['flt', 'fge', 'feq', 'fne',
        ('bcsel', 'a', (op, 'd', 'b'), (op, 'd', 'c'))),
    ]
 
+# This section contains "late" optimizations that should be run before
+# creating ffmas and calling regular optimizations for the final time.
+# Optimizations should go here if they help code generation and conflict
+# with the regular optimizations.
+before_ffma_optimizations = [
+   # Propagate constants down multiplication chains
+   (('~fmul(is_used_once)', ('fmul(is_used_once)', 'a(is_not_const)', '#b'), 'c(is_not_const)'), ('fmul', ('fmul', a, c), b)),
+   (('imul(is_used_once)', ('imul(is_used_once)', 'a(is_not_const)', '#b'), 'c(is_not_const)'), ('imul', ('imul', a, c), b)),
+   (('~fadd(is_used_once)', ('fadd(is_used_once)', 'a(is_not_const)', '#b'), 'c(is_not_const)'), ('fadd', ('fadd', a, c), b)),
+   (('iadd(is_used_once)', ('iadd(is_used_once)', 'a(is_not_const)', '#b'), 'c(is_not_const)'), ('iadd', ('iadd', a, c), b)),
+
+   (('~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),
+   (('iadd', ('ineg', a), ('iadd', a, b)), b),
+   (('iadd', a, ('iadd', ('ineg', a), b)), b),
+   (('~fadd', ('fneg', a), ('fadd', a, b)), b),
+   (('~fadd', a, ('fadd', ('fneg', a), b)), b),
+]
+
 # 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
@@ -556,5 +577,7 @@ late_optimizations = [
 ]
 
 print nir_algebraic.AlgebraicPass("nir_opt_algebraic", optimizations).render()
+print nir_algebraic.AlgebraicPass("nir_opt_algebraic_before_ffma",
+                                  before_ffma_optimizations).render()
 print nir_algebraic.AlgebraicPass("nir_opt_algebraic_late",
                                   late_optimizations).render()

diff --git a/src/intel/compiler/brw_nir.c b/src/intel/compiler/brw_nir.c
index 36ccdf3cb1eb39ac32596bfcaa3940803718ce16..3c0a7ced572706e58a0ca9b515f763fff05a93d2 100644 (file)
--- a/src/intel/compiler/brw_nir.c
+++ b/src/intel/compiler/brw_nir.c
@@ -605,6 +605,12 @@ brw_postprocess_nir(nir_shader *nir, const struct brw_compiler *compiler,
 
    UNUSED bool progress; /* Written by OPT */
 
+
+   do {
+      progress = false;
+      OPT(nir_opt_algebraic_before_ffma);
+   } while (progress);
+
    nir = nir_optimize(nir, compiler, is_scalar);
 
    if (devinfo->gen >= 6) {