diff options
Diffstat (limited to 'Eigen/src/Core/arch/GPU/TypeCasting.h')
-rw-r--r-- | Eigen/src/Core/arch/GPU/TypeCasting.h | 161 |
1 files changed, 0 insertions, 161 deletions
diff --git a/Eigen/src/Core/arch/GPU/TypeCasting.h b/Eigen/src/Core/arch/GPU/TypeCasting.h index 57a55d08b..c278f3fe8 100644 --- a/Eigen/src/Core/arch/GPU/TypeCasting.h +++ b/Eigen/src/Core/arch/GPU/TypeCasting.h @@ -14,64 +14,6 @@ namespace Eigen { namespace internal { -template<> -struct scalar_cast_op<float, Eigen::half> { - EIGEN_EMPTY_STRUCT_CTOR(scalar_cast_op) - typedef Eigen::half result_type; - EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Eigen::half operator() (const float& a) const { - #if (defined(EIGEN_HAS_CUDA_FP16) && defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 300) || \ - (defined(EIGEN_HAS_HIP_FP16) && defined(EIGEN_HIP_DEVICE_COMPILE)) - return __float2half(a); - #else - return Eigen::half(a); - #endif - } -}; - -template<> -struct functor_traits<scalar_cast_op<float, Eigen::half> > -{ enum { Cost = NumTraits<float>::AddCost, PacketAccess = false }; }; - - -template<> -struct scalar_cast_op<int, Eigen::half> { - EIGEN_EMPTY_STRUCT_CTOR(scalar_cast_op) - typedef Eigen::half result_type; - EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE Eigen::half operator() (const int& a) const { - #if (defined(EIGEN_HAS_CUDA_FP16) && defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 300) || \ - (defined(EIGEN_HAS_HIP_FP16) && defined(EIGEN_HIP_DEVICE_COMPILE)) - return __float2half(static_cast<float>(a)); - #else - return Eigen::half(static_cast<float>(a)); - #endif - } -}; - -template<> -struct functor_traits<scalar_cast_op<int, Eigen::half> > -{ enum { Cost = NumTraits<float>::AddCost, PacketAccess = false }; }; - - -template<> -struct scalar_cast_op<Eigen::half, float> { - EIGEN_EMPTY_STRUCT_CTOR(scalar_cast_op) - typedef float result_type; - EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE float operator() (const Eigen::half& a) const { - #if (defined(EIGEN_HAS_CUDA_FP16) && defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 300) || \ - (defined(EIGEN_HAS_HIP_FP16) && defined(EIGEN_HIP_DEVICE_COMPILE)) - return __half2float(a); - #else - return static_cast<float>(a); - #endif - } -}; - -template<> -struct functor_traits<scalar_cast_op<Eigen::half, float> > -{ enum { Cost = NumTraits<float>::AddCost, PacketAccess = false }; }; - - - #if (defined(EIGEN_HAS_CUDA_FP16) && defined(EIGEN_CUDA_ARCH) && EIGEN_CUDA_ARCH >= 300) || \ (defined(EIGEN_HAS_HIP_FP16) && defined(EIGEN_HIP_DEVICE_COMPILE)) @@ -104,109 +46,6 @@ template<> EIGEN_DEVICE_FUNC EIGEN_STRONG_INLINE half2 pcast<float4, half2>(cons return __floats2half2_rn(a.x, a.y); } -#elif defined EIGEN_VECTORIZE_AVX512 -template <> -struct type_casting_traits<half, float> { - enum { - VectorizedCast = 1, - SrcCoeffRatio = 1, - TgtCoeffRatio = 1 - }; -}; - -template<> EIGEN_STRONG_INLINE Packet16f pcast<Packet16h, Packet16f>(const Packet16h& a) { - return half2float(a); -} - -template <> -struct type_casting_traits<float, half> { - enum { - VectorizedCast = 1, - SrcCoeffRatio = 1, - TgtCoeffRatio = 1 - }; -}; - -template<> EIGEN_STRONG_INLINE Packet16h pcast<Packet16f, Packet16h>(const Packet16f& a) { - return float2half(a); -} - -#elif defined EIGEN_VECTORIZE_AVX - -template <> -struct type_casting_traits<Eigen::half, float> { - enum { - VectorizedCast = 1, - SrcCoeffRatio = 1, - TgtCoeffRatio = 1 - }; -}; - -template<> EIGEN_STRONG_INLINE Packet8f pcast<Packet8h, Packet8f>(const Packet8h& a) { - return half2float(a); -} - -template <> -struct type_casting_traits<float, Eigen::half> { - enum { - VectorizedCast = 1, - SrcCoeffRatio = 1, - TgtCoeffRatio = 1 - }; -}; - -template<> EIGEN_STRONG_INLINE Packet8h pcast<Packet8f, Packet8h>(const Packet8f& a) { - return float2half(a); -} - -// Disable the following code since it's broken on too many platforms / compilers. -//#elif defined(EIGEN_VECTORIZE_SSE) && (!EIGEN_ARCH_x86_64) && (!EIGEN_COMP_MSVC) -#elif 0 - -template <> -struct type_casting_traits<Eigen::half, float> { - enum { - VectorizedCast = 1, - SrcCoeffRatio = 1, - TgtCoeffRatio = 1 - }; -}; - -template<> EIGEN_STRONG_INLINE Packet4f pcast<Packet4h, Packet4f>(const Packet4h& a) { - __int64_t a64 = _mm_cvtm64_si64(a.x); - Eigen::half h = raw_uint16_to_half(static_cast<unsigned short>(a64)); - float f1 = static_cast<float>(h); - h = raw_uint16_to_half(static_cast<unsigned short>(a64 >> 16)); - float f2 = static_cast<float>(h); - h = raw_uint16_to_half(static_cast<unsigned short>(a64 >> 32)); - float f3 = static_cast<float>(h); - h = raw_uint16_to_half(static_cast<unsigned short>(a64 >> 48)); - float f4 = static_cast<float>(h); - return _mm_set_ps(f4, f3, f2, f1); -} - -template <> -struct type_casting_traits<float, Eigen::half> { - enum { - VectorizedCast = 1, - SrcCoeffRatio = 1, - TgtCoeffRatio = 1 - }; -}; - -template<> EIGEN_STRONG_INLINE Packet4h pcast<Packet4f, Packet4h>(const Packet4f& a) { - EIGEN_ALIGN16 float aux[4]; - pstore(aux, a); - Eigen::half h0(aux[0]); - Eigen::half h1(aux[1]); - Eigen::half h2(aux[2]); - Eigen::half h3(aux[3]); - - Packet4h result; - result.x = _mm_set_pi16(h3.x, h2.x, h1.x, h0.x); - return result; -} - #endif } // end namespace internal |