From: Rob Clark <robcl...@freedesktop.org>
In freedreno these get implemented as the matching f* instruction plus a
u2f to convert the result to float 1.0/0.0. But less lines of code to
just let nir_opt_algebraic handle this for us, plus opens up some small
window for other opt passes to improve (ie. if some shader ended up with
both a flt and slt with same src args, for example).
Signed-off-by: Rob Clark <robcl...@freedesktop.org>
---
src/glsl/nir/nir.h | 3 +++
src/glsl/nir/nir_opt_algebraic.py | 5 +++++
2 files changed, 8 insertions(+)
diff --git a/src/glsl/nir/nir.h b/src/glsl/nir/nir.h
index 669a26e..11505f9 100644
--- a/src/glsl/nir/nir.h
+++ b/src/glsl/nir/nir.h
@@ -1371,6 +1371,9 @@ typedef struct nir_shader_compiler_options {
/** lowers fneg and ineg to fsub and isub. */
bool lower_negate;
+ /* lower {slt,sge,seq,sne} to {flt,fge,feq,fne} + u2f: */
+ bool lower_scmp;
+
/**
* Does the driver support real 32-bit integers? (Otherwise, integers
* are simulated by floats.)
diff --git a/src/glsl/nir/nir_opt_algebraic.py
b/src/glsl/nir/nir_opt_algebraic.py
index ef855aa..6bd4187 100644
--- a/src/glsl/nir/nir_opt_algebraic.py
+++ b/src/glsl/nir/nir_opt_algebraic.py
@@ -95,6 +95,11 @@ optimizations = [
(('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)),
+ (('slt', a, b), ('u2f', ('flt', a, b)), 'options->lower_scmp'),
+ (('sge', a, b), ('u2f', ('fge', a, b)), 'options->lower_scmp'),
+ (('seq', a, b), ('u2f', ('feq', a, b)), 'options->lower_scmp'),
+ (('sne', a, b), ('u2f', ('fne', a, b)), 'options->lower_scmp'),
+
# Comparison with the same args. Note that these are not done for
# the float versions because NaN always returns false on float
# inequalities.
--
2.1.0
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
