Mesa (master): nir: Correctly constant fold fsign(NaN) and fsign(-0)

[GitLab Mirror]

Module: Mesa
Branch: master
Commit: 71961c73a94ceca9dbeb8718acbe2c3c7a7ee717
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=71961c73a94ceca9dbeb8718acbe2c3c7a7ee717

Author: Ian Romanick <ian.d.romanick at intel.com>
Date:   Tue Feb 18 12:34:15 2020 -0800

nir: Correctly constant fold fsign(NaN) and fsign(-0)

GLSL and SPIR-V GLSL.std.450 don't have any requirements for fsign(NaN),
and both only require that FSign(-0.0) == 0.0.  OpenCL, on the other
hand, requires sign(-0.0) be exactly -0.0.  It also requires that
sign(NaN) be exactly 0.0.

In practice, this change is difficult to test.  Our GLSL frontend
already constant folds sign(NaN) to 0.0 before even getting to NIR.  As
far as I can tell, glslang does the same.  I don't have a good way to
run an OpenCL SPIR-V test.  Maybe SPIR-V GLSL.std.450 assembly?

No shader-db or fossil-db changes on any Intel platform.

Acked-by: Rhys Perry <pendingchaos02 at gmail.com>
Reviewed-by: Caio Marcelo de Oliveira Filho <caio.oliveira at intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6358>

---

 src/compiler/nir/nir_opcodes.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/src/compiler/nir/nir_opcodes.py b/src/compiler/nir/nir_opcodes.py
index bc0c7fe2c0b..6a16e280d0c 100644
--- a/src/compiler/nir/nir_opcodes.py
+++ b/src/compiler/nir/nir_opcodes.py
@@ -208,11 +208,11 @@ unop("inot", tint, "~src0") # invert every bit of the integer
 # number, and the result is not +0.0, the result should definitely **not** be
 # NaN.
 #
-# fsign(NaN) = (False ? 0.0 : (False ? 1.0 : -1.0) = -1.0.  This is allowed by
-# the spec, but it is not the preferred value.
+# The values returned for constant folding match the behavior required by
+# OpenCL.
 unop("fsign", tfloat, ("bit_size == 64 ? " +
-                       "((src0 == 0.0) ? 0.0 : ((src0 > 0.0) ? 1.0 : -1.0)) : " +
-                       "((src0 == 0.0f) ? 0.0f : ((src0 > 0.0f) ? 1.0f : -1.0f))"))
+                       "(isnan(src0) ? 0.0  : ((src0 == 0.0 ) ? src0 : (src0 > 0.0 ) ? 1.0  : -1.0 )) : " +
+                       "(isnan(src0) ? 0.0f : ((src0 == 0.0f) ? src0 : (src0 > 0.0f) ? 1.0f : -1.0f))"))
 unop("isign", tint, "(src0 == 0) ? 0 : ((src0 > 0) ? 1 : -1)")
 unop("iabs", tint, "(src0 < 0) ? -src0 : src0")
 unop("fabs", tfloat, "fabs(src0)")