High precision implementation for sin/cos/tan
This implementation allows all dEQP precision test to pass
for sin, cos and tan function tests.
Change-Id: I33a24497dea68ab2de2e65931f50f2dd4298523c
Reviewed-on: https://swiftshader-review.googlesource.com/13555
Reviewed-by: Nicolas Capens <nicolascapens@google.com>
Tested-by: Alexis Hétu <sugoi@google.com>
diff --git a/src/OpenGL/compiler/OutputASM.cpp b/src/OpenGL/compiler/OutputASM.cpp
index 889f44e..f9a4dbb 100644
--- a/src/OpenGL/compiler/OutputASM.cpp
+++ b/src/OpenGL/compiler/OutputASM.cpp
@@ -1894,6 +1894,12 @@
instruction->dst.integer = (dst->getBasicType() == EbtInt);
}
+ if(src0)
+ {
+ TIntermTyped* src = src0->getAsTyped();
+ instruction->dst.partialPrecision = src && (src->getPrecision() <= EbpLow);
+ }
+
argument(instruction->src[0], src0, index0);
argument(instruction->src[1], src1, index1);
argument(instruction->src[2], src2, index2);
diff --git a/src/Shader/ShaderCore.cpp b/src/Shader/ShaderCore.cpp
index 53644b8..017d0f8 100644
--- a/src/Shader/ShaderCore.cpp
+++ b/src/Shader/ShaderCore.cpp
@@ -287,6 +287,21 @@
Float4 y = x * Float4(1.59154943e-1f); // 1/2pi
y = y - Round(y);
+ if(!pp)
+ {
+ // From the paper: "A Fast, Vectorizable Algorithm for Producing Single-Precision Sine-Cosine Pairs"
+ // This implementation passes OpenGL ES 3.0 precision requirements, at the cost of more operations:
+ // !pp : 17 mul, 7 add, 1 sub, 1 reciprocal
+ // pp : 4 mul, 2 add, 2 abs
+
+ Float4 y2 = y * y;
+ Float4 c1 = y2 * (y2 * (y2 * Float4(-0.0204391631f) + Float4(0.2536086171f)) + Float4(-1.2336977925f)) + Float4(1.0f);
+ Float4 s1 = y * (y2 * (y2 * (y2 * Float4(-0.0046075748f) + Float4(0.0796819754f)) + Float4(-0.645963615f)) + Float4(1.5707963235f));
+ Float4 c2 = (c1 * c1) - (s1 * s1);
+ Float4 s2 = Float4(2.0f) * s1 * c1;
+ return Float4(2.0f) * s2 * c2 * reciprocal(s2 * s2 + c2 * c2, pp, true);
+ }
+
const Float4 A = Float4(-16.0f);
const Float4 B = Float4(8.0f);
const Float4 C = Float4(7.75160950e-1f);
