From: Ian Romanick <ian.d.romanick@intel.com>
Date: Thu, 22 Feb 2018 02:30:20 +0000 (-0800)
Subject: nir/algebraic: Recognize open-coded copysign(1.0, a)
X-Git-Url: https://git.libre-soc.org/?a=commitdiff_plain;h=25bfba3335d0fb2822dac02822d096dd75b3ef4d;p=mesa.git

nir/algebraic: Recognize open-coded copysign(1.0, a)

All of the affected shaders are in Mad Max.  The inner part of the
pattern is itself an open-coded sign(a).  I tried using that as a
pattern, but the results were not good.  A bunch of shaders were helped
for instructions, but overall cycles, spill, and fills were hurt.

v2: Rebase on 1-bit Boolean changes.

v3: Fix order of copysign() parameters in comments and commit message.
Noticed by Matt.

All Gen7+ platforms had similar results. (Skylake shown)
total instructions in shared programs: 15282141 -> 15282073 (<.01%)
instructions in affected programs: 6106 -> 6038 (-1.11%)
helped: 17
HURT: 0
helped stats (abs) min: 4 max: 4 x̄: 4.00 x̃: 4
helped stats (rel) min: 1.02% max: 2.20% x̄: 1.15% x̃: 1.06%
95% mean confidence interval for instructions value: -4.00 -4.00
95% mean confidence interval for instructions %-change: -1.30% -1.00%
Instructions are helped.

total cycles in shared programs: 372597886 -> 372595954 (<.01%)
cycles in affected programs: 32701 -> 30769 (-5.91%)
helped: 17
HURT: 0
helped stats (abs) min: 6 max: 216 x̄: 113.65 x̃: 118
helped stats (rel) min: 0.40% max: 21.86% x̄: 6.20% x̃: 5.83%
95% mean confidence interval for cycles value: -152.84 -74.45
95% mean confidence interval for cycles %-change: -8.89% -3.51%
Cycles are helped.

No changes on any Gen6 or earlier platforms.

Reviewed-by: Matt Turner <mattst88@gmail.com>
---

diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py
index e8696abaf52..f275d70d276 100644
--- a/src/compiler/nir/nir_opt_algebraic.py
+++ b/src/compiler/nir/nir_opt_algebraic.py
@@ -394,6 +394,12 @@ optimizations = [
    (('ior', ('uge', 1, a), ('ieq', a, 2)), ('uge', 2, a)),
    (('ior', ('uge', 2, a), ('ieq', a, 3)), ('uge', 3, a)),
 
+   # The (i2f32, ...) part is an open-coded fsign.  When that is combined with
+   # the bcsel, it's basically copysign(1.0, a).  There is no copysign in NIR,
+   # so emit an open-coded version of that.
+   (('bcsel@32', ('feq', a, 0.0), 1.0, ('i2f32', ('iadd', ('b2i32', ('flt', 0.0, 'a@32')), ('ineg', ('b2i32', ('flt', 'a@32', 0.0)))))),
+    ('ior', 0x3f800000, ('iand', a, 0x80000000))),
+
    (('ior', a, ('ieq', a, False)), True),
    (('ior', a, ('inot', a)), -1),