[Mesa-dev] [PATCH 6/9] nir: Add a flag for lowering ffma.

[Eric Anholt]

vc4 cse/algebraic-disabled stats:
total uniforms in shared programs: 13966 -> 13791 (-1.25%)
uniforms in affected programs:     435 -> 260 (-40.23%)
total instructions in shared programs: 44732 -> 44356 (-0.84%)
instructions in affected programs:     9599 -> 9223 (-3.92%)
---
 src/gallium/drivers/vc4/vc4_program.c | 1 +
 src/glsl/nir/nir.h                    | 1 +
 src/glsl/nir/nir_opt_algebraic.py     | 3 ++-
 3 files changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/gallium/drivers/vc4/vc4_program.c b/src/gallium/drivers/vc4/vc4_program.c
index 304cd04..8be2d15 100644
--- a/src/gallium/drivers/vc4/vc4_program.c
+++ b/src/gallium/drivers/vc4/vc4_program.c
@@ -2098,6 +2098,7 @@ nir_to_qir(struct vc4_compile *c)
 }
 
 static const nir_shader_compiler_options nir_options = {
+        .lower_ffma = true,
         .lower_fpow = true,
         .lower_fsqrt = true,
         .lower_negate = true,
diff --git a/src/glsl/nir/nir.h b/src/glsl/nir/nir.h
index 0746adf..f7f5d33 100644
--- a/src/glsl/nir/nir.h
+++ b/src/glsl/nir/nir.h
@@ -1327,6 +1327,7 @@ typedef struct nir_function {
                   exec_list_get_head(&(func)->overload_list), node)
 
 typedef struct nir_shader_compiler_options {
+   bool lower_ffma;
    bool lower_fpow;
    bool lower_fsqrt;
    /** lowers fneg and ineg to fsub and isub. */
diff --git a/src/glsl/nir/nir_opt_algebraic.py b/src/glsl/nir/nir_opt_algebraic.py
index 7d5f543..8a14fe0 100644
--- a/src/glsl/nir/nir_opt_algebraic.py
+++ b/src/glsl/nir/nir_opt_algebraic.py
@@ -71,7 +71,8 @@ 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)),
+   (('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)),
-- 
2.1.4