diff options
| author | Jonathan Marek <jonathan@marek.ca> | 2019-05-08 10:26:49 -0400 |
|---|---|---|
| committer | Jonathan Marek <jonathan@marek.ca> | 2019-07-24 17:36:21 -0400 |
| commit | bc3b6168bac39a16326c730a9d0ae97b45c7df23 (patch) | |
| tree | 58816c0efaa72880f75d457c725ff8b417ecf188 /src/compiler/nir/nir_opt_algebraic.py | |
| parent | 5a4e71c082886810504ecfa329fb57050acc623f (diff) | |
| download | mesa-bc3b6168bac39a16326c730a9d0ae97b45c7df23.tar.gz | |
nir: replace lower_sincos with algebraic opt
This version has less ops for the same precision.
Signed-off-by: Jonathan Marek <jonathan@marek.ca>
Reviewed-by: Vasily Khoruzhick <anarsoul@gmail.com>
Acked-by: Matt Turner <mattst88@gmail.com>
Diffstat (limited to 'src/compiler/nir/nir_opt_algebraic.py')
| -rw-r--r-- | src/compiler/nir/nir_opt_algebraic.py | 10 |
1 files changed, 10 insertions, 0 deletions
diff --git a/src/compiler/nir/nir_opt_algebraic.py b/src/compiler/nir/nir_opt_algebraic.py index 00145fc80f5..951771fa37a 100644 --- a/src/compiler/nir/nir_opt_algebraic.py +++ b/src/compiler/nir/nir_opt_algebraic.py @@ -29,6 +29,7 @@ from collections import OrderedDict import nir_algebraic from nir_opcodes import type_sizes import itertools +from math import pi # Convenience variables a = 'a' @@ -74,6 +75,12 @@ e = 'e' # another condition, the two can be separated by a comma (e.g., # "(many-comm-expr,is_used_once)"). +# based on https://web.archive.org/web/20180105155939/http://forum.devmaster.net/t/fast-and-accurate-sine-cosine/9648 +def lowered_sincos(c): + x = ('fsub', ('fmul', 2.0, ('ffract', ('fadd', ('fmul', 0.5 / pi, a), c))), 1.0) + x = ('fmul', ('fsub', x, ('fmul', x, ('fabs', x))), 4.0) + return ('ffma', ('ffma', x, ('fabs', x), ('fneg', x)), 0.225, x) + optimizations = [ (('imul', a, '#b@32(is_pos_power_of_two)'), ('ishl', a, ('find_lsb', b)), '!options->lower_bitshift'), @@ -645,6 +652,9 @@ optimizations = [ (('~frcp', ('fsqrt', a)), ('frsq', a)), (('fsqrt', a), ('frcp', ('frsq', a)), 'options->lower_fsqrt'), (('~frcp', ('frsq', a)), ('fsqrt', a), '!options->lower_fsqrt'), + # Trig + (('fsin', a), lowered_sincos(0.5), 'options->lower_sincos'), + (('fcos', a), lowered_sincos(0.75), 'options->lower_sincos'), # Boolean simplifications (('i2b32(is_used_by_if)', a), ('ine32', a, 0)), (('i2b1(is_used_by_if)', a), ('ine', a, 0)), |