diff options
| author |  Benoit Steiner <benoit.steiner.goog@gmail.com> | 2015-02-26 09:42:41 -0800 |
|---|---|---|
| committer | Benoit Steiner <benoit.steiner.goog@gmail.com> | 2015-02-26 09:42:41 -0800 |
| commit | f41b1f1666e91dc674a42fed9c444c91f483133f (patch) | |
| tree | f81dfac2e8e4ec0c3038cb8a814de31a7dafb9b6 /Eigen/src/Core/arch/SSE/MathFunctions.h | |
| parent | 8e817b65d02e6f6c2d4a0d0085212db0abe6c485 (diff) | |
Added support for fast reciprocal square root computation.
Diffstat (limited to 'Eigen/src/Core/arch/SSE/MathFunctions.h')
| -rw-r--r-- | Eigen/src/Core/arch/SSE/MathFunctions.h | 18 |
1 files changed, 16 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 |