aboutsummaryrefslogtreecommitdiffhomepage
path: root/Eigen/src/Core/arch/SSE
diff options
context:
space:
mode:
Diffstat (limited to 'Eigen/src/Core/arch/SSE')
-rw-r--r--Eigen/src/Core/arch/SSE/MathFunctions.h18
-rwxr-xr-xEigen/src/Core/arch/SSE/PacketMath.h2
2 files changed, 18 insertions, 2 deletions
diff --git a/Eigen/src/Core/arch/SSE/MathFunctions.h b/Eigen/src/Core/arch/SSE/MathFunctions.h
index f86c0a39a..f9cb93bfc 100644
--- a/Eigen/src/Core/arch/SSE/MathFunctions.h
+++ b/Eigen/src/Core/arch/SSE/MathFunctions.h
@@ -462,11 +462,25 @@ Packet4f psqrt<Packet4f>(const Packet4f& _x)
#else
-template<> EIGEN_STRONG_INLINE Packet4f psqrt<Packet4f>(const Packet4f& x) { return _mm_sqrt_ps(x); }
+template<>EIGEN_DEFINE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS EIGEN_UNUSED
+Packet4f psqrt<Packet4f>(const Packet4f& x) { return _mm_sqrt_ps(x); }
#endif
-template<> EIGEN_STRONG_INLINE Packet2d psqrt<Packet2d>(const Packet2d& x) { return _mm_sqrt_pd(x); }
+template<> EIGEN_DEFINE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS EIGEN_UNUSED
+Packet2d psqrt<Packet2d>(const Packet2d& x) { return _mm_sqrt_pd(x); }
+
+template<> EIGEN_DEFINE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS EIGEN_UNUSED
+Packet4f prsqrt<Packet4f>(const Packet4f& x) {
+ // Unfortunately we can't use the much faster mm_rqsrt_ps since it only provides an approximation.
+ return _mm_div_ps(pset1<Packet4f>(1.0f), _mm_sqrt_ps(x));
+}
+
+template<> EIGEN_DEFINE_FUNCTION_ALLOWING_MULTIPLE_DEFINITIONS EIGEN_UNUSED
+Packet2d prsqrt<Packet2d>(const Packet2d& x) {
+ // Unfortunately we can't use the much faster mm_rqsrt_pd since it only provides an approximation.
+ return _mm_div_pd(pset1<Packet2d>(1.0), _mm_sqrt_pd(x));
+}
} // end namespace internal
diff --git a/Eigen/src/Core/arch/SSE/PacketMath.h b/Eigen/src/Core/arch/SSE/PacketMath.h
index 3653783fd..38a84273d 100755
--- a/Eigen/src/Core/arch/SSE/PacketMath.h
+++ b/Eigen/src/Core/arch/SSE/PacketMath.h
@@ -108,6 +108,7 @@ template<> struct packet_traits<float> : default_packet_traits
HasLog = 1,
HasExp = 1,
HasSqrt = 1,
+ HasRsqrt = 1,
HasBlend = 1
};
};
@@ -124,6 +125,7 @@ template<> struct packet_traits<double> : default_packet_traits
HasDiv = 1,
HasExp = 1,
HasSqrt = 1,
+ HasRsqrt = 1,
HasBlend = 1
};
};