From d2447bb039c268aa21a5ca358cc2d91abe4f4d21 Mon Sep 17 00:00:00 2001 From: Sang-Hoon Park Date: Mon, 18 Jan 2021 09:41:37 +0000 Subject: Decouple data types of elementwise kernels Partially implements: COMPMID-4003 Change-Id: Ie51e43e24fb9a6b5b96d13cdc3d72fbda027a68b Signed-off-by: Sang-Hoon Park Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4873 Tested-by: Arm Jenkins Reviewed-by: Georgios Pinitas Comments-Addressed: Arm Jenkins --- .../NEON/kernels/NEElementwiseOperationKernel.cpp | 1187 ++------------------ .../kernels/elementwise/impl/elementwise_list.h | 486 ++++++++ .../elementwise/impl/elementwise_quantized_list.h | 654 +++++++++++ 3 files changed, 1209 insertions(+), 1118 deletions(-) create mode 100644 src/core/NEON/kernels/elementwise/impl/elementwise_list.h create mode 100644 src/core/NEON/kernels/elementwise/impl/elementwise_quantized_list.h diff --git a/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp b/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp index 4d67ec3986..b250465e14 100644 --- a/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp +++ b/src/core/NEON/kernels/NEElementwiseOperationKernel.cpp @@ -26,1180 +26,131 @@ #include "arm_compute/core/Helpers.h" #include "arm_compute/core/IAccessWindow.h" #include "src/core/CPP/Validate.h" -#include "src/core/NEON/NEAsymm.h" -#include "src/core/NEON/NEFixedPoint.h" -#include "src/core/NEON/wrapper/wrapper.h" +#include "src/core/NEON/kernels/elementwise/impl/elementwise_list.h" +#include "src/core/NEON/kernels/elementwise/impl/elementwise_quantized_list.h" #include "src/core/SVE/kernels/elementwise/impl/elementwise_list.h" #include "src/core/SVE/kernels/elementwise/impl/elementwise_quantized_list.h" +#include "src/core/common/Registrars.h" #include "src/core/helpers/AutoConfiguration.h" #include "src/core/helpers/WindowHelpers.h" #include -#include namespace arm_compute { namespace { -float32x4x4_t load_quantized(const uint8_t *input1_ptr, const int32x4_t &offset, const float32x4_t &scale) +using ElementwiseSelector = std::add_pointer::type; +using UKernelType = NEElementwiseOperationKernel::ElementwiseFunction; +struct ElementwiseKernel { - qasymm8x16_t x = vld1q_u8(input1_ptr); - const float32x4x4_t out = - { - { - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_low_u16(vmovl_u8(vget_low_u8(x))))), offset)), scale), - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_high_u16(vmovl_u8(vget_low_u8(x))))), offset)), scale), - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_low_u16(vmovl_u8(vget_high_u8(x))))), offset)), scale), - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_high_u16(vmovl_u8(vget_high_u8(x))))), offset)), scale), - } - }; - return out; -} - -float32x4x4_t load_quantized_signed(const int8_t *input1_ptr, const int32x4_t &offset, const float32x4_t &scale) -{ - qasymm8x16_signed_t x = vld1q_s8(input1_ptr); - const float32x4x4_t out = - { - { - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_low_s16(vmovl_s8(vget_low_s8(x)))), offset)), scale), - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_high_s16(vmovl_s8(vget_low_s8(x)))), offset)), scale), - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_low_s16(vmovl_s8(vget_high_s8(x)))), offset)), scale), - vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_high_s16(vmovl_s8(vget_high_s8(x)))), offset)), scale), - } - }; - return out; -} - -void store_quantized(uint8_t *output_ptr, const uint32x4x4_t &out) -{ - const uint8x8_t pa = vqmovn_u16(vcombine_u16(vqmovn_u32(out.val[0]), vqmovn_u32(out.val[1]))); - const uint8x8_t pb = vqmovn_u16(vcombine_u16(vqmovn_u32(out.val[2]), vqmovn_u32(out.val[3]))); - vst1q_u8(output_ptr, vcombine_u8(pa, pb)); -} - -void store_quantized(uint8_t *output_ptr, const int32x4x4_t &out) -{ - const uint8x8_t pa = vqmovun_s16(vcombine_s16(vqmovn_s32(out.val[0]), vqmovn_s32(out.val[1]))); - const uint8x8_t pb = vqmovun_s16(vcombine_s16(vqmovn_s32(out.val[2]), vqmovn_s32(out.val[3]))); - vst1q_u8(output_ptr, vcombine_u8(pa, pb)); -} - -void store_quantized(uint8_t *output_ptr, const float32x4x4_t &rf, const float32x4_t &offset, const float32x4_t &invscale) -{ - int32x4x4_t out = - { - { - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[0], invscale)), - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[1], invscale)), - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[2], invscale)), - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[3], invscale)), - } - }; - store_quantized(output_ptr, out); -} - -void store_quantized_signed(int8_t *output_ptr, const int32x4x4_t &out) -{ - const int8x8_t pa = vqmovn_s16(vcombine_s16(vqmovn_s32(out.val[0]), vqmovn_s32(out.val[1]))); - const int8x8_t pb = vqmovn_s16(vcombine_s16(vqmovn_s32(out.val[2]), vqmovn_s32(out.val[3]))); - vst1q_s8(output_ptr, vcombine_s8(pa, pb)); -} - -void store_quantized_signed(int8_t *output_ptr, const float32x4x4_t &rf, const float32x4_t &offset, const float32x4_t &invscale) -{ - int32x4x4_t out = - { - { - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[0], invscale)), - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[1], invscale)), - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[2], invscale)), - vcvtq_s32_f32(vmlaq_f32(offset, rf.val[3], invscale)), - } - }; - store_quantized_signed(output_ptr, out); -} - -template -inline ScalarType elementwise_arithm_op_scalar(const ScalarType &a, const ScalarType &b) -{ - auto res = ScalarType(0); - - switch(op) - { - case ArithmeticOperation::MAX: - res = std::max(a, b); - break; - case ArithmeticOperation::MIN: - res = std::min(a, b); - break; - case ArithmeticOperation::SQUARED_DIFF: - { - res = (a - b) * (a - b); - break; - } - case ArithmeticOperation::PRELU: - { - res = (a > 0 ? a : a * b); - break; - } - case ArithmeticOperation::DIV: - { - res = a / b; - if(std::is_integral::value) - { - res = (b == 0) ? 0 : res; - if(static_cast(a) % static_cast(b) != 0 && ((a < 0) != (b < 0))) - { - --res; - } - } - break; - } - case ArithmeticOperation::POWER: - { - res = std::pow(a, b); - break; - } - default: - ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); - } - return res; -} - -template -inline uint8_t elementwise_arithm_op_quantized_scalar(const float &a, const float &b, UniformQuantizationInfo qinfo) -{ - return quantize_qasymm8(elementwise_arithm_op_scalar(a, b), qinfo); -} - -template -inline int8_t elementwise_arithm_op_quantized_signed_scalar(const float &a, const float &b, UniformQuantizationInfo qinfo) -{ - return quantize_qasymm8_signed(elementwise_arithm_op_scalar(a, b), qinfo); -} - -template -inline typename VectorType::type elementwise_arithm_op(const typename VectorType::type &a, const typename VectorType::type &b) -{ - using vec_type = typename VectorType::type; - using scalar_type = typename VectorType::scalar_type; - using tag_type = typename VectorType::tag_type; - - vec_type res = wrapper::vdup_n(static_cast(0), tag_type{}); - - switch(op) - { - case ArithmeticOperation::MAX: - res = wrapper::vmax(a, b); - break; - case ArithmeticOperation::MIN: - res = wrapper::vmin(a, b); - break; - case ArithmeticOperation::SQUARED_DIFF: - { - const vec_type tmp = wrapper::vsub(a, b); - res = wrapper::vmul(tmp, tmp); - break; - } - case ArithmeticOperation::PRELU: - { - const vec_type zero = wrapper::vdup_n(static_cast(0), tag_type{}); - const vec_type tmp = wrapper::vmul(a, b); - const auto gt = wrapper::vcgt(a, zero); - - res = wrapper::vbsl(gt, a, tmp); - break; - } - - default: - ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); - } - - return res; -} - -template <> -inline int32x4_t elementwise_arithm_op>(const int32x4_t &a, const int32x4_t &b) -{ - return vcvtq_s32_f32(vfloorq_f32(wrapper::vdiv(vcvtq_f32_s32(a), vcvtq_f32_s32(b)))); -} - -template <> -inline float32x4_t elementwise_arithm_op>(const float32x4_t &a, const float32x4_t &b) -{ - return wrapper::vdiv(a, b); -} - -template <> -inline float32x4_t elementwise_arithm_op>(const float32x4_t &a, const float32x4_t &b) -{ - return wrapper::vpow(a, b); -} - -#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC -template <> -inline float16x8_t elementwise_arithm_op>(const float16x8_t &a, const float16x8_t &b) -{ - return wrapper::vdiv(a, b); -} - -template <> -inline float16x8_t elementwise_arithm_op>(const float16x8_t &a, const float16x8_t &b) -{ - return wrapper::vpow(a, b); -} -#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC - -template -inline float32x4x4_t elementwise_arithm_op(const float32x4x4_t &a, const float32x4x4_t &b) -{ - using neon_vector_float = wrapper::traits::neon_vector; - float32x4x4_t out = - { - { - elementwise_arithm_op(a.val[0], b.val[0]), - elementwise_arithm_op(a.val[1], b.val[1]), - elementwise_arithm_op(a.val[2], b.val[2]), - elementwise_arithm_op(a.val[3], b.val[3]), - } - }; - return out; -} - -template -inline typename VectorType::type elementwise_arithm_op_broadcast(const typename VectorType::type &a, const ScalarType &broadcast_value, const bool reorder) -{ - using tag_type = typename VectorType::tag_type; - using vec_type = typename VectorType::type; - - vec_type broadcast_vector = wrapper::vdup_n(broadcast_value, tag_type{}); - return elementwise_arithm_op(reorder ? broadcast_vector : a, reorder ? a : broadcast_vector); -} - -template -inline uint8_t elementwise_comp_op_scalar(const InputScalarType &a, const InputScalarType &b) -{ - bool res = false; - - switch(op) - { - case ComparisonOperation::Equal: - res = (a == b); - break; - case ComparisonOperation::NotEqual: - res = (a != b); - break; - case ComparisonOperation::Greater: - res = (a > b); - break; - case ComparisonOperation::GreaterEqual: - res = (a >= b); - break; - case ComparisonOperation::Less: - res = (a < b); - break; - case ComparisonOperation::LessEqual: - res = (a <= b); - break; - default: - ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); - } - return res ? ~static_cast(0) : static_cast(0); -} - -template -inline uint8_t elementwise_comp_op_quantized_scalar(const float &a, const float &b, UniformQuantizationInfo qinfo) -{ - ARM_COMPUTE_UNUSED(qinfo); - return elementwise_comp_op_scalar(a, b); -} - -template -inline OutputVectorType elementwise_comp_op(const InputVectorType &a, const InputVectorType &b) -{ - OutputVectorType res = { 0, 0, 0, 0 }; - - switch(op) - { - case ComparisonOperation::Equal: - res = wrapper::vceq(a, b); - break; - case ComparisonOperation::NotEqual: - res = wrapper::vnot(wrapper::vceq(a, b)); - break; - case ComparisonOperation::Greater: - res = wrapper::vcgt(a, b); - break; - case ComparisonOperation::GreaterEqual: - res = wrapper::vcge(a, b); - break; - case ComparisonOperation::Less: - res = wrapper::vcgt(b, a); - break; - case ComparisonOperation::LessEqual: - res = wrapper::vcge(b, a); - break; - default: - ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); - } - - return res; -} + const char *name; + const ElementwiseSelector is_selected; + UKernelType *ukernel; +}; -template -inline uint32x4x4_t elementwise_comp_op(const float32x4x4_t &a, const float32x4x4_t &b) +template +inline bool is_selected(DataType data_type) { - uint32x4x4_t out = - { - { - elementwise_comp_op(a.val[0], b.val[0]), - elementwise_comp_op(a.val[1], b.val[1]), - elementwise_comp_op(a.val[2], b.val[2]), - elementwise_comp_op(a.val[3], b.val[3]) - } - }; - return out; + return dt == data_type; } -template -inline OutputVectorType elementwise_comp_op_broadcast(const InputVectorType &a, const InputScalarType &broadcast_value, const bool reorder) +template +static ElementwiseKernel generate_kernel(UKernelType *ukernel) { - InputVectorType broadcast_vector = wrapper::vdup_n(broadcast_value, wrapper::traits::vector_128_tag()); - return elementwise_comp_op(reorder ? broadcast_vector : a, reorder ? a : broadcast_vector); -} + std::string kernel_name("op_"); + kernel_name += string_from_data_type(input_data_type) + "_"; + kernel_name += string_from_data_type(input_data_type) + "_"; + kernel_name += string_from_data_type(output_data_type); -template -inline int elementwise_arithm_op_loop(int window_start_x, int window_end_x, int window_step_x, - const ScalarType *input1_ptr, const ScalarType *input2_ptr, ScalarType *output_ptr) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const auto a = wrapper::vloadq(input1_ptr + x); - const auto b = wrapper::vloadq(input2_ptr + x); - wrapper::vstore(output_ptr + x, elementwise_arithm_op(a, b)); - } - return x; -} - -template -inline int elementwise_arithm_op_quantized_loop(int window_start_x, int window_end_x, int window_step_x, - const uint8_t *input1_ptr, const uint8_t *input2_ptr, uint8_t *output_ptr, - int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, - float32x4_t voffseto, float32x4_t invvscaleo) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - // Get inputs and compute output - const float32x4x4_t af = load_quantized(input1_ptr + x, voffset1, vscale1); - const float32x4x4_t bf = load_quantized(input2_ptr + x, voffset2, vscale2); - const float32x4x4_t rf = elementwise_arithm_op(af, bf); - store_quantized(output_ptr + x, rf, voffseto, invvscaleo); - } - return x; -} - -template -inline int elementwise_arithm_op_quantized_singed_loop(int window_start_x, int window_end_x, int window_step_x, - const int8_t *input1_ptr, const int8_t *input2_ptr, int8_t *output_ptr, - int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, - float32x4_t voffseto, float32x4_t invvscaleo) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - // Get inputs and compute output - const float32x4x4_t af = load_quantized_signed(input1_ptr + x, voffset1, vscale1); - const float32x4x4_t bf = load_quantized_signed(input2_ptr + x, voffset2, vscale2); - const float32x4x4_t rf = elementwise_arithm_op(af, bf); - store_quantized_signed(output_ptr + x, rf, voffseto, invvscaleo); - } - return x; -} - -template -inline int elementwise_arithm_op_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, - const ScalarType *non_broadcast_input_ptr, const ScalarType &broadcast_value, ScalarType *output_ptr, const bool reorder) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const auto a = wrapper::vloadq((non_broadcast_input_ptr + x)); - wrapper::vstore(output_ptr + x, elementwise_arithm_op_broadcast(a, broadcast_value, reorder)); - } - return x; -} - -template -inline int elementwise_arithm_op_quantized_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, - const uint8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, uint8_t *output_ptr, - int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, - float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const float32x4x4_t af = load_quantized(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); - const float32x4x4_t rf = elementwise_arithm_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); - store_quantized(output_ptr + x, rf, voffseto, invvscaleo); - } - return x; -} -template -inline int elementwise_arithm_op_quantized_signed_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, - const int8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, int8_t *output_ptr, - int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, - float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const float32x4x4_t af = load_quantized_signed(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); - const float32x4x4_t rf = elementwise_arithm_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); - store_quantized_signed(output_ptr + x, rf, voffseto, invvscaleo); - } - return x; -} - -template -inline int elementwise_comp_op_8_loop(int window_start_x, int window_end_x, int window_step_x, - const InputScalarType *input1_ptr, const InputScalarType *input2_ptr, uint8_t *output_ptr) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const auto a = wrapper::vloadq(input1_ptr + x); - const auto b = wrapper::vloadq(input2_ptr + x); - const auto res = elementwise_comp_op(a, b); - wrapper::vstore(output_ptr + x, res); - } - return x; -} - -template -inline int elementwise_comp_op_16_loop(int window_start_x, int window_end_x, int window_step_x, - const InputScalarType *input1_ptr, const InputScalarType *input2_ptr, uint8_t *output_ptr) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const auto a = wrapper::vloadq(input1_ptr + x); - const auto b = wrapper::vloadq(input2_ptr + x); - const auto res = elementwise_comp_op(a, b); - wrapper::vstore(output_ptr + x, wrapper::vmovn(res)); - } - return x; -} - -template -inline int elementwise_comp_op_32_loop(int window_start_x, int window_end_x, int window_step_x, - const InputScalarType *input1_ptr, const InputScalarType *input2_ptr, uint8_t *output_ptr) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - auto a = wrapper::vloadq(input1_ptr + x); - auto b = wrapper::vloadq(input2_ptr + x); - const auto res = elementwise_comp_op(a, b); - a = wrapper::vloadq(input1_ptr + x + 4); - b = wrapper::vloadq(input2_ptr + x + 4); - const auto res2 = elementwise_comp_op(a, b); - wrapper::vstore(output_ptr + x, wrapper::vmovn(wrapper::vcombine(wrapper::vmovn(res), wrapper::vmovn(res2)))); - } - if(x <= window_end_x - 4) - { - const auto a = wrapper::vloadq(input1_ptr + x); - const auto b = wrapper::vloadq(input2_ptr + x); - const auto res = elementwise_comp_op(a, b); - for(int i = 0; i < 4; i++) - { - *(output_ptr + x + i) = wrapper::vgetlane(res, i); - } - x = +4; - } - return x; -} - -template -inline int elementwise_comp_op_quantized_loop(int window_start_x, int window_end_x, int window_step_x, - const uint8_t *input1_ptr, const uint8_t *input2_ptr, uint8_t *output_ptr, - int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, - float32x4_t voffseto, float32x4_t invvscaleo) -{ - ARM_COMPUTE_UNUSED(voffseto, invvscaleo); - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const float32x4x4_t af = load_quantized(input1_ptr + x, voffset1, vscale1); - const float32x4x4_t bf = load_quantized(input2_ptr + x, voffset2, vscale2); - const uint32x4x4_t rf = elementwise_comp_op(af, bf); - store_quantized(output_ptr + x, rf); - } - return x; -} - -template -inline int elementwise_comp_op_quantized_signed_loop(int window_start_x, int window_end_x, int window_step_x, - const int8_t *input1_ptr, const int8_t *input2_ptr, uint8_t *output_ptr, - int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, - float32x4_t voffseto, float32x4_t invvscaleo) -{ - ARM_COMPUTE_UNUSED(voffseto, invvscaleo); - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const float32x4x4_t af = load_quantized_signed(input1_ptr + x, voffset1, vscale1); - const float32x4x4_t bf = load_quantized_signed(input2_ptr + x, voffset2, vscale2); - const uint32x4x4_t rf = elementwise_comp_op(af, bf); - store_quantized(output_ptr + x, rf); - } - return x; -} - -template -inline int elementwise_comp_op_broadcast_8_loop(int window_start_x, int window_end_x, int window_step_x, - const InputScalarType *non_broadcast_input_ptr, const InputScalarType &broadcast_value, uint8_t *output_ptr, const bool reorder) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const auto a = elementwise_comp_op_broadcast(wrapper::vloadq((non_broadcast_input_ptr + x)), broadcast_value, reorder); - wrapper::vstore(output_ptr + x, a); - } - return x; -} - -template -inline int elementwise_comp_op_broadcast_16_loop(int window_start_x, int window_end_x, int window_step_x, - const InputScalarType *non_broadcast_input_ptr, const InputScalarType &broadcast_value, uint8_t *output_ptr, const bool reorder) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const auto a = elementwise_comp_op_broadcast(wrapper::vloadq((non_broadcast_input_ptr + x)), broadcast_value, reorder); - wrapper::vstore(output_ptr + x, wrapper::vmovn(a)); - } - return x; -} - -template -inline int elementwise_comp_op_broadcast_32_loop(int window_start_x, int window_end_x, int window_step_x, - const InputScalarType *non_broadcast_input_ptr, const InputScalarType &broadcast_value, uint8_t *output_ptr, const bool reorder) -{ - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const auto a = elementwise_comp_op_broadcast(wrapper::vloadq(non_broadcast_input_ptr + x), broadcast_value, reorder); - const auto b = elementwise_comp_op_broadcast(wrapper::vloadq(non_broadcast_input_ptr + x + 4), broadcast_value, reorder); - wrapper::vstore(output_ptr + x, wrapper::vmovn(wrapper::vcombine(wrapper::vmovn(a), wrapper::vmovn(b)))); - } - if(x <= window_end_x - 4) - { - const auto a = elementwise_comp_op_broadcast(wrapper::vloadq((non_broadcast_input_ptr + x)), broadcast_value, reorder); - for(int i = 0; i < 4; i++) - { - *(output_ptr + x + i) = wrapper::vgetlane(a, i); - } - x = +4; - } - return x; -} - -template -inline int elementwise_comp_op_quantized_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, - const uint8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, uint8_t *output_ptr, - int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, - float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) -{ - ARM_COMPUTE_UNUSED(voffseto, invvscaleo); - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const float32x4x4_t af = load_quantized(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); - const uint32x4x4_t rf = elementwise_comp_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); - store_quantized(output_ptr + x, rf); - } - return x; -} - -template -inline int elementwise_comp_op_quantized_signed_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, - const int8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, uint8_t *output_ptr, - int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, - float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) -{ - ARM_COMPUTE_UNUSED(voffseto, invvscaleo); - int x = window_start_x; - for(; x <= (window_end_x - window_step_x); x += window_step_x) - { - const float32x4x4_t af = load_quantized_signed(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); - const uint32x4x4_t rf = elementwise_comp_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); - store_quantized(output_ptr + x, rf); - } - return x; -} - -template -void elementwise_op(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, - OutputScalarType (*scalar_func)(const InputScalarType &, const InputScalarType &), - int (*broadcast_func)(int, int, int, const InputScalarType *, const InputScalarType &, OutputScalarType *, const bool), - int (*neon_func)(int, int, int, const InputScalarType *, const InputScalarType *, OutputScalarType *)) -{ - // Create input windows - Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); - Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); - - // Clear X Dimension on execution window as we handle manually - Window win = window; - win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - const int window_step_x = std::min(16 / static_cast(sizeof(OutputScalarType)), 8); - const auto window_start_x = static_cast(window.x().start()); - const auto window_end_x = static_cast(window.x().end()); - const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); - - if(is_broadcast_across_x) - { - const bool is_broadcast_input_2 = input2_win.x().step() == 0; - Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; - Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; - const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; - const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; - - // Clear X Dimension on execution window as we handle manually - non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator broadcast_input(broadcast_tensor, broadcast_win); - Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - auto output_ptr = reinterpret_cast(output.ptr()); - const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); - const InputScalarType broadcast_value = *reinterpret_cast(broadcast_input.ptr()); - - int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_value, output_ptr, !is_broadcast_input_2); - for(; x < window_end_x; ++x) - { - const auto a = *(non_broadcast_input_ptr + x); - *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? broadcast_value : a, !is_broadcast_input_2 ? a : broadcast_value); - } - }, - broadcast_input, non_broadcast_input, output); - } - else - { - // Clear X Dimension on execution window as we handle manually - input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator input1(in1, input1_win); - Iterator input2(in2, input2_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - auto output_ptr = reinterpret_cast(output.ptr()); - const auto input1_ptr = reinterpret_cast(input1.ptr()); - const auto input2_ptr = reinterpret_cast(input2.ptr()); - - int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr); - for(; x < window_end_x; ++x) - { - const auto a = *(input1_ptr + x); - const auto b = *(input2_ptr + x); - *(output_ptr + x) = (*scalar_func)(a, b); - } - }, - input1, input2, output); - } -} - -#if !defined(__ARM_FEATURE_SVE2) -void elementwise_op_quantized(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, - uint8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), - int (*broadcast_func)(int, int, int, const uint8_t *, float32x4x4_t, uint8_t *, int32x4_t, float32x4_t, - float32x4_t, float32x4_t, const bool), - int (*neon_func)(int, int, int, const uint8_t *, const uint8_t *, uint8_t *, - int32x4_t, int32x4_t, float32x4_t, float32x4_t, - float32x4_t, float32x4_t)) -{ - // Create input windows - Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); - Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); - - // Clear X Dimension on execution window as we handle manually - Window win = window; - win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - const int window_step_x = 16; - const auto window_start_x = static_cast(window.x().start()); - const auto window_end_x = static_cast(window.x().end()); - const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); - - const UniformQuantizationInfo output_qinfo = out->info()->quantization_info().uniform(); - - // Output quantization info (add 0.5 to round toward the nearest integer - 0.5 rounds away from zero) - const float32x4_t voffseto = vdupq_n_f32(output_qinfo.offset + 0.5f); - const float32x4_t invvscaleo = vdupq_n_f32(1.f / output_qinfo.scale); - - if(is_broadcast_across_x) - { - // Select the broadcast input on the X axis - const bool is_broadcast_input_2 = input2_win.x().step() == 0; - Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; - Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; - const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; - const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; - - const UniformQuantizationInfo broadcast_qinfo = broadcast_tensor->info()->quantization_info().uniform(); - const UniformQuantizationInfo non_broadcast_qinfo = non_broadcast_tensor->info()->quantization_info().uniform(); - - const int32x4_t voffset_non_broadcast = vdupq_n_s32(non_broadcast_qinfo.offset); - const float32x4_t vscale_non_broadcast = vdupq_n_f32(non_broadcast_qinfo.scale); - - // Clear X Dimension on execution window as we handle manually - non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator broadcast_input(broadcast_tensor, broadcast_win); - Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); - const auto output_ptr = reinterpret_cast(output.ptr()); - - const uint8_t broadcast_value = *reinterpret_cast(broadcast_input.ptr()); - const float32x4x4_t broadcast_vector = vdequantize(vdupq_n_u8(broadcast_value), broadcast_qinfo); - - int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_vector, output_ptr, - voffset_non_broadcast, vscale_non_broadcast, voffseto, invvscaleo, !is_broadcast_input_2); - for(; x < window_end_x; ++x) - { - const float afs = dequantize_qasymm8(*(non_broadcast_input_ptr + x), non_broadcast_qinfo); - const float bfs = dequantize_qasymm8(broadcast_value, broadcast_qinfo); - *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? bfs : afs, !is_broadcast_input_2 ? afs : bfs, output_qinfo); - } - }, - broadcast_input, non_broadcast_input, output); - } - else - { - const UniformQuantizationInfo input1_qinfo = in1->info()->quantization_info().uniform(); - const UniformQuantizationInfo input2_qinfo = in2->info()->quantization_info().uniform(); - - // Input1 quantization info - const int32x4_t voffset1 = vdupq_n_s32(input1_qinfo.offset); - const float32x4_t vscale1 = vdupq_n_f32(input1_qinfo.scale); - - // Input2 quantization info - const int32x4_t voffset2 = vdupq_n_s32(input2_qinfo.offset); - const float32x4_t vscale2 = vdupq_n_f32(input2_qinfo.scale); - - // Clear X Dimension on execution window as we handle manually - input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator input1(in1, input1_win); - Iterator input2(in2, input2_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - const auto input1_ptr = reinterpret_cast(input1.ptr()); - const auto input2_ptr = reinterpret_cast(input2.ptr()); - const auto output_ptr = reinterpret_cast(output.ptr()); - - int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr, voffset1, voffset2, - vscale1, vscale2, voffseto, invvscaleo); - for(; x < window_end_x; ++x) - { - const float afs = dequantize_qasymm8(*(input1_ptr + x), input1_qinfo); - const float bfs = dequantize_qasymm8(*(input2_ptr + x), input2_qinfo); - *(output_ptr + x) = (*scalar_func)(afs, bfs, output_qinfo); - } - }, - input1, input2, output); - } -} - -void elementwise_comp_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, - uint8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), - int (*broadcast_func)(int, int, int, const int8_t *, float32x4x4_t, uint8_t *, int32x4_t, float32x4_t, - float32x4_t, float32x4_t, const bool), - int (*neon_func)(int, int, int, const int8_t *, const int8_t *, uint8_t *, - int32x4_t, int32x4_t, float32x4_t, float32x4_t, - float32x4_t, float32x4_t)) -{ - // Create input windows - Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); - Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); - - // Clear X Dimension on execution window as we handle manually - Window win = window; - win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - const int window_step_x = 16; - const auto window_start_x = static_cast(window.x().start()); - const auto window_end_x = static_cast(window.x().end()); - const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); - - const UniformQuantizationInfo output_qinfo = out->info()->quantization_info().uniform(); - - const float32x4_t voffseto = vdupq_n_f32(output_qinfo.offset); - const float32x4_t invvscaleo = vdupq_n_f32(1.f / output_qinfo.scale); - - if(is_broadcast_across_x) - { - // Select the broadcast input on the X axis - const bool is_broadcast_input_2 = input2_win.x().step() == 0; - Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; - Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; - const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; - const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; - - const UniformQuantizationInfo broadcast_qinfo = broadcast_tensor->info()->quantization_info().uniform(); - const UniformQuantizationInfo non_broadcast_qinfo = non_broadcast_tensor->info()->quantization_info().uniform(); - - const int32x4_t voffset_non_broadcast = vdupq_n_s32(non_broadcast_qinfo.offset); - const float32x4_t vscale_non_broadcast = vdupq_n_f32(non_broadcast_qinfo.scale); - - // Clear X Dimension on execution window as we handle manually - non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator broadcast_input(broadcast_tensor, broadcast_win); - Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); - const auto output_ptr = reinterpret_cast(output.ptr()); - - const int8_t broadcast_value = *reinterpret_cast(broadcast_input.ptr()); - const float32x4x4_t broadcast_vector = vdequantize(vdupq_n_s8(broadcast_value), broadcast_qinfo); - - int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_vector, output_ptr, - voffset_non_broadcast, vscale_non_broadcast, voffseto, invvscaleo, !is_broadcast_input_2); - for(; x < window_end_x; ++x) - { - const float afs = dequantize_qasymm8_signed(*(non_broadcast_input_ptr + x), non_broadcast_qinfo); - const float bfs = dequantize_qasymm8_signed(broadcast_value, broadcast_qinfo); - *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? bfs : afs, !is_broadcast_input_2 ? afs : bfs, output_qinfo); - } - }, - broadcast_input, non_broadcast_input, output); - } - else - { - const UniformQuantizationInfo input1_qinfo = in1->info()->quantization_info().uniform(); - const UniformQuantizationInfo input2_qinfo = in2->info()->quantization_info().uniform(); - - // Input1 quantization info - const int32x4_t voffset1 = vdupq_n_s32(input1_qinfo.offset); - const float32x4_t vscale1 = vdupq_n_f32(input1_qinfo.scale); - - // Input2 quantization info - const int32x4_t voffset2 = vdupq_n_s32(input2_qinfo.offset); - const float32x4_t vscale2 = vdupq_n_f32(input2_qinfo.scale); - - // Clear X Dimension on execution window as we handle manually - input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator input1(in1, input1_win); - Iterator input2(in2, input2_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - const auto input1_ptr = reinterpret_cast(input1.ptr()); - const auto input2_ptr = reinterpret_cast(input2.ptr()); - const auto output_ptr = reinterpret_cast(output.ptr()); - - int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr, voffset1, voffset2, - vscale1, vscale2, voffseto, invvscaleo); - for(; x < window_end_x; ++x) - { - const float afs = dequantize_qasymm8_signed(*(input1_ptr + x), input1_qinfo); - const float bfs = dequantize_qasymm8_signed(*(input2_ptr + x), input2_qinfo); - *(output_ptr + x) = (*scalar_func)(afs, bfs, output_qinfo); - } - }, - input1, input2, output); - } -} - -void elementwise_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, - int8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), - int (*broadcast_func)(int, int, int, const int8_t *, float32x4x4_t, int8_t *, int32x4_t, float32x4_t, - float32x4_t, float32x4_t, const bool), - int (*neon_func)(int, int, int, const int8_t *, const int8_t *, int8_t *, - int32x4_t, int32x4_t, float32x4_t, float32x4_t, - float32x4_t, float32x4_t)) -{ - // Create input windows - Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); - Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); - - // Clear X Dimension on execution window as we handle manually - Window win = window; - win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - const int window_step_x = 16; - const auto window_start_x = static_cast(window.x().start()); - const auto window_end_x = static_cast(window.x().end()); - const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); - - const UniformQuantizationInfo output_qinfo = out->info()->quantization_info().uniform(); - - const float32x4_t voffseto = vdupq_n_f32(output_qinfo.offset); - const float32x4_t invvscaleo = vdupq_n_f32(1.f / output_qinfo.scale); - - if(is_broadcast_across_x) - { - // Select the broadcast input on the X axis - const bool is_broadcast_input_2 = input2_win.x().step() == 0; - Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; - Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; - const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; - const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; - - const UniformQuantizationInfo broadcast_qinfo = broadcast_tensor->info()->quantization_info().uniform(); - const UniformQuantizationInfo non_broadcast_qinfo = non_broadcast_tensor->info()->quantization_info().uniform(); - - const int32x4_t voffset_non_broadcast = vdupq_n_s32(non_broadcast_qinfo.offset); - const float32x4_t vscale_non_broadcast = vdupq_n_f32(non_broadcast_qinfo.scale); - - // Clear X Dimension on execution window as we handle manually - non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator broadcast_input(broadcast_tensor, broadcast_win); - Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); - const auto output_ptr = reinterpret_cast(output.ptr()); - - const int8_t broadcast_value = *reinterpret_cast(broadcast_input.ptr()); - const float32x4x4_t broadcast_vector = vdequantize(vdupq_n_s8(broadcast_value), broadcast_qinfo); - - int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_vector, output_ptr, - voffset_non_broadcast, vscale_non_broadcast, voffseto, invvscaleo, !is_broadcast_input_2); - for(; x < window_end_x; ++x) - { - const float afs = dequantize_qasymm8_signed(*(non_broadcast_input_ptr + x), non_broadcast_qinfo); - const float bfs = dequantize_qasymm8_signed(broadcast_value, broadcast_qinfo); - *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? bfs : afs, !is_broadcast_input_2 ? afs : bfs, output_qinfo); - } - }, - broadcast_input, non_broadcast_input, output); - } - else - { - const UniformQuantizationInfo input1_qinfo = in1->info()->quantization_info().uniform(); - const UniformQuantizationInfo input2_qinfo = in2->info()->quantization_info().uniform(); - - // Input1 quantization info - const int32x4_t voffset1 = vdupq_n_s32(input1_qinfo.offset); - const float32x4_t vscale1 = vdupq_n_f32(input1_qinfo.scale); - - // Input2 quantization info - const int32x4_t voffset2 = vdupq_n_s32(input2_qinfo.offset); - const float32x4_t vscale2 = vdupq_n_f32(input2_qinfo.scale); - - // Clear X Dimension on execution window as we handle manually - input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); - - Iterator input1(in1, input1_win); - Iterator input2(in2, input2_win); - Iterator output(out, win); - - execute_window_loop(win, [&](const Coordinates &) - { - const auto input1_ptr = reinterpret_cast(input1.ptr()); - const auto input2_ptr = reinterpret_cast(input2.ptr()); - const auto output_ptr = reinterpret_cast(output.ptr()); - - int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr, voffset1, voffset2, - vscale1, vscale2, voffseto, invvscaleo); - for(; x < window_end_x; ++x) - { - const float afs = dequantize_qasymm8_signed(*(input1_ptr + x), input1_qinfo); - const float bfs = dequantize_qasymm8_signed(*(input2_ptr + x), input2_qinfo); - *(output_ptr + x) = (*scalar_func)(afs, bfs, output_qinfo); - } - }, - input1, input2, output); - } -} -#endif /* !defined(__ARM_FEATURE_SVE2) */ - -template -void elementwise_comp_op_8(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - elementwise_op(in1, in2, out, window, - &elementwise_comp_op_scalar, - &elementwise_comp_op_broadcast_8_loop, - &elementwise_comp_op_8_loop); -} - -template -void elementwise_comp_op_16(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - elementwise_op(in1, in2, out, window, - &elementwise_comp_op_scalar, - &elementwise_comp_op_broadcast_16_loop, - &elementwise_comp_op_16_loop); -} - -template -void elementwise_comp_op_32(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - elementwise_op(in1, in2, out, window, - &elementwise_comp_op_scalar, - &elementwise_comp_op_broadcast_32_loop, - &elementwise_comp_op_32_loop); -} - -template -void elementwise_arithm_op(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - using scalar_type = typename VectorType::scalar_type; - - elementwise_op(in1, in2, out, window, - &elementwise_arithm_op_scalar, - &elementwise_arithm_op_broadcast_loop, - &elementwise_arithm_op_loop); -} - -template -void elementwise_arithm_op_quantized(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - elementwise_op_quantized(in1, in2, out, window, &elementwise_arithm_op_quantized_scalar, - &elementwise_arithm_op_quantized_broadcast_loop, - &elementwise_arithm_op_quantized_loop); -} -template -void elementwise_arithm_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - elementwise_op_quantized_signed(in1, in2, out, window, &elementwise_arithm_op_quantized_signed_scalar, - &elementwise_arithm_op_quantized_signed_broadcast_loop, - &elementwise_arithm_op_quantized_singed_loop); -} - -template -void elementwise_comp_op_quantized(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - elementwise_op_quantized(in1, in2, out, window, &elementwise_comp_op_quantized_scalar, - &elementwise_comp_op_quantized_broadcast_loop, - &elementwise_comp_op_quantized_loop); -} - -template -void elementwise_comp_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) -{ - elementwise_comp_quantized_signed(in1, in2, out, window, &elementwise_comp_op_quantized_scalar, - &elementwise_comp_op_quantized_signed_broadcast_loop, - &elementwise_comp_op_quantized_signed_loop); -} - -std::function -configure_func(const ITensorInfo *input1, const ITensorInfo *input2, ITensorInfo *output, - std::map map_function) -{ - std::string function_to_call("op_"); - function_to_call += string_from_data_type(input1->data_type()) + "_"; - function_to_call += string_from_data_type(input2->data_type()) + "_"; - function_to_call += string_from_data_type(output->data_type()); - - auto it = map_function.find(function_to_call); - - if(it != map_function.end()) - { - auto func = it->second; - return [func](const ITensor * input1, const ITensor * input2, ITensor * output, const Window & window) - { - func(input1, input2, output, window); - }; - } - return nullptr; + return { kernel_name.c_str(), is_selected, ukernel }; } template std::function configure_arithm_func(const ITensorInfo *input1, const ITensorInfo *input2, ITensorInfo *output) { - static std::map map_function = + ARM_COMPUTE_UNUSED(input2, output); + static ElementwiseKernel kernels[] = { #if defined(__ARM_FEATURE_SVE) - { "op_F32_F32_F32", &arm_compute::cpu::sve::elementwise_arithmetic_op }, - { "op_S32_S32_S32", &arm_compute::cpu::sve::elementwise_arithmetic_op }, + generate_kernel(REGISTER_FP32_SVE((arm_compute::cpu::sve::elementwise_arithmetic_op))), + generate_kernel(REGISTER_INTEGER_SVE((arm_compute::cpu::sve::elementwise_arithmetic_op))), #else /* defined(__ARM_FEATURE_SVE) */ - { "op_F32_F32_F32", &elementwise_arithm_op> }, - { "op_S32_S32_S32", &elementwise_arithm_op> }, + generate_kernel(REGISTER_FP32_NEON((arm_compute::cpu::elementwise_arithm_op>))), + generate_kernel(REGISTER_INTEGER_NEON((arm_compute::cpu::elementwise_arithm_op>))), #endif /* defined(__ARM_FEATURE_SVE) */ #if defined(__ARM_FEATURE_SVE2) - { "op_QASYMM8_QASYMM8_QASYMM8", &arm_compute::cpu::sve::elementwise_arithmetic_quantized_op }, - { "op_QASYMM8_SIGNED_QASYMM8_SIGNED_QASYMM8_SIGNED", &arm_compute::cpu::sve::elementwise_arithmetic_quantized_op }, + generate_kernel(REGISTER_QASYMM8_SVE((arm_compute::cpu::sve::elementwise_arithmetic_quantized_op))), + generate_kernel(REGISTER_QASYMM8_SIGNED_SVE((arm_compute::cpu::sve::elementwise_arithmetic_quantized_op))), #else /* defined(__ARM_FEATURE_SVE2) */ - { "op_QASYMM8_QASYMM8_QASYMM8", &elementwise_arithm_op_quantized }, - { "op_QASYMM8_SIGNED_QASYMM8_SIGNED_QASYMM8_SIGNED", &elementwise_arithm_op_quantized_signed }, + generate_kernel(REGISTER_QASYMM8_NEON((arm_compute::cpu::elementwise_arithm_op_quantized))), + generate_kernel(REGISTER_QASYMM8_SIGNED_NEON((arm_compute::cpu::elementwise_arithm_op_quantized_signed))), #endif /* defined(__ARM_FEATURE_SVE2) */ - { "op_S16_S16_S16", &elementwise_arithm_op> }, - }; #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC #if defined(__ARM_FEATURE_SVE) - map_function["op_F16_F16_F16"] = &arm_compute::cpu::sve::elementwise_arithmetic_op; + generate_kernel(REGISTER_FP16_SVE((arm_compute::cpu::sve::elementwise_arithmetic_op))), #else /* defined(__ARM_FEATURE_SVE) */ - map_function["op_F16_F16_F16"] = &elementwise_arithm_op>; + generate_kernel(REGISTER_FP16_NEON((arm_compute::cpu::elementwise_arithm_op>))), #endif /* defined(__ARM_FEATURE_SVE) */ #endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */ + generate_kernel(REGISTER_INTEGER_NEON((arm_compute::cpu::elementwise_arithm_op>))), + }; + + for(const auto &uk : kernels) + { + if(uk.is_selected(input1->data_type())) + { + return uk.ukernel; + } + } - return configure_func(input1, input2, output, map_function); + return nullptr; } template std::function configure_comp_func(const ITensorInfo *input1, const ITensorInfo *input2, ITensorInfo *output) { - static std::map map_function = + ARM_COMPUTE_UNUSED(input2, output); + static ElementwiseKernel kernels[] = { #if defined(__ARM_FEATURE_SVE) - { "op_U8_U8_U8", &arm_compute::cpu::sve::elementwise_comparison_op }, - { "op_F32_F32_U8", &arm_compute::cpu::sve::elementwise_comparison_op }, - { "op_S16_S16_U8", &arm_compute::cpu::sve::elementwise_comparison_op }, - { "op_S32_S32_U8", &arm_compute::cpu::sve::elementwise_comparison_op }, + generate_kernel(REGISTER_INTEGER_SVE((arm_compute::cpu::sve::elementwise_comparison_op))), + generate_kernel(REGISTER_FP32_SVE((arm_compute::cpu::sve::elementwise_comparison_op))), + generate_kernel(REGISTER_INTEGER_SVE((arm_compute::cpu::sve::elementwise_comparison_op))), + generate_kernel(REGISTER_INTEGER_SVE((arm_compute::cpu::sve::elementwise_comparison_op))), #else /* defined(__ARM_FEATURE_SVE) */ - { "op_U8_U8_U8", &elementwise_comp_op_8 }, - { "op_F32_F32_U8", &elementwise_comp_op_32 }, - { "op_S16_S16_U8", &elementwise_comp_op_16 }, - { "op_S32_S32_U8", &elementwise_comp_op_32 }, + generate_kernel(REGISTER_INTEGER_NEON((arm_compute::cpu::elementwise_comp_op_8))), + generate_kernel(REGISTER_FP32_NEON((arm_compute::cpu::elementwise_comp_op_32))), + generate_kernel(REGISTER_INTEGER_NEON((arm_compute::cpu::elementwise_comp_op_16))), + generate_kernel(REGISTER_INTEGER_NEON((arm_compute::cpu::elementwise_comp_op_32))), #endif /* defined(__ARM_FEATURE_SVE) */ #if defined(__ARM_FEATURE_SVE2) - { "op_QASYMM8_SIGNED_QASYMM8_SIGNED_U8", &arm_compute::cpu::sve::elementwise_comparison_quantized_op }, - { "op_QASYMM8_QASYMM8_U8", &arm_compute::cpu::sve::elementwise_comparison_quantized_op } + generate_kernel(REGISTER_QASYMM8_SIGNED_SVE((arm_compute::cpu::sve::elementwise_comparison_quantized_op))), + generate_kernel(REGISTER_QASYMM8_SVE((arm_compute::cpu::sve::elementwise_comparison_quantized_op))), #else /* defined(__ARM_FEATURE_SVE2) */ - { "op_QASYMM8_SIGNED_QASYMM8_SIGNED_U8", &elementwise_comp_op_quantized_signed }, - { "op_QASYMM8_QASYMM8_U8", &elementwise_comp_op_quantized } + generate_kernel(REGISTER_QASYMM8_SIGNED_NEON((arm_compute::cpu::elementwise_comp_op_quantized_signed))), + generate_kernel(REGISTER_QASYMM8_NEON((arm_compute::cpu::elementwise_comp_op_quantized))), #endif /* defined(__ARM_FEATURE_SVE2) */ - }; #ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC #if defined(__ARM_FEATURE_SVE) - map_function["op_F16_F16_U8"] = &arm_compute::cpu::sve::elementwise_comparison_op; + generate_kernel(REGISTER_FP16_SVE((arm_compute::cpu::sve::elementwise_comparison_op))), #else /* defined(__ARM_FEATURE_SVE) */ - map_function["op_F16_F16_U8"] = &elementwise_comp_op_16; + generate_kernel(REGISTER_FP16_NEON((arm_compute::cpu::elementwise_comp_op_16))), #endif /* defined(__ARM_FEATURE_SVE) */ #endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */ + }; - return configure_func(input1, input2, output, map_function); + for(const auto &uk : kernels) + { + if(uk.is_selected(input1->data_type())) + { + return uk.ukernel; + } + } + + return nullptr; } } // namespace diff --git a/src/core/NEON/kernels/elementwise/impl/elementwise_list.h b/src/core/NEON/kernels/elementwise/impl/elementwise_list.h new file mode 100644 index 0000000000..43e44be5e2 --- /dev/null +++ b/src/core/NEON/kernels/elementwise/impl/elementwise_list.h @@ -0,0 +1,486 @@ +/* + * Copyright (c) 2021 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef SRC_CORE_NEON_KERNELS_ELEMENTWISE_LIST_H +#define SRC_CORE_NEON_KERNELS_ELEMENTWISE_LIST_H + +#include "src/core/NEON/NEAsymm.h" +#include "src/core/NEON/wrapper/wrapper.h" +#include "src/core/helpers/WindowHelpers.h" + +namespace arm_compute +{ +namespace cpu +{ +template +void elementwise_op(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, + OutputScalarType (*scalar_func)(const InputScalarType &, const InputScalarType &), + int (*broadcast_func)(int, int, int, const InputScalarType *, const InputScalarType &, OutputScalarType *, const bool), + int (*neon_func)(int, int, int, const InputScalarType *, const InputScalarType *, OutputScalarType *)) +{ + // Create input windows + Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); + Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); + + // Clear X Dimension on execution window as we handle manually + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + const int window_step_x = std::min(16 / static_cast(sizeof(OutputScalarType)), 8); + const auto window_start_x = static_cast(window.x().start()); + const auto window_end_x = static_cast(window.x().end()); + const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); + + if(is_broadcast_across_x) + { + const bool is_broadcast_input_2 = input2_win.x().step() == 0; + Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; + Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; + const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; + const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; + + // Clear X Dimension on execution window as we handle manually + non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator broadcast_input(broadcast_tensor, broadcast_win); + Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + auto output_ptr = reinterpret_cast(output.ptr()); + const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); + const InputScalarType broadcast_value = *reinterpret_cast(broadcast_input.ptr()); + + int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_value, output_ptr, !is_broadcast_input_2); + for(; x < window_end_x; ++x) + { + const auto a = *(non_broadcast_input_ptr + x); + *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? broadcast_value : a, !is_broadcast_input_2 ? a : broadcast_value); + } + }, + broadcast_input, non_broadcast_input, output); + } + else + { + // Clear X Dimension on execution window as we handle manually + input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator input1(in1, input1_win); + Iterator input2(in2, input2_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + auto output_ptr = reinterpret_cast(output.ptr()); + const auto input1_ptr = reinterpret_cast(input1.ptr()); + const auto input2_ptr = reinterpret_cast(input2.ptr()); + + int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr); + for(; x < window_end_x; ++x) + { + const auto a = *(input1_ptr + x); + const auto b = *(input2_ptr + x); + *(output_ptr + x) = (*scalar_func)(a, b); + } + }, + input1, input2, output); + } +} + +template +inline ScalarType elementwise_arithm_op_scalar(const ScalarType &a, const ScalarType &b) +{ + auto res = ScalarType(0); + + switch(op) + { + case ArithmeticOperation::MAX: + res = std::max(a, b); + break; + case ArithmeticOperation::MIN: + res = std::min(a, b); + break; + case ArithmeticOperation::SQUARED_DIFF: + { + res = (a - b) * (a - b); + break; + } + case ArithmeticOperation::PRELU: + { + res = (a > 0 ? a : a * b); + break; + } + case ArithmeticOperation::DIV: + { + res = a / b; + if(std::is_integral::value) + { + res = (b == 0) ? 0 : res; + if(static_cast(a) % static_cast(b) != 0 && ((a < 0) != (b < 0))) + { + --res; + } + } + break; + } + case ArithmeticOperation::POWER: + { + res = std::pow(a, b); + break; + } + default: + ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); + } + return res; +} + +template +inline typename VectorType::type elementwise_arithm_op(const typename VectorType::type &a, const typename VectorType::type &b) +{ + using vec_type = typename VectorType::type; + using scalar_type = typename VectorType::scalar_type; + using tag_type = typename VectorType::tag_type; + + vec_type res = wrapper::vdup_n(static_cast(0), tag_type{}); + + switch(op) + { + case ArithmeticOperation::MAX: + res = wrapper::vmax(a, b); + break; + case ArithmeticOperation::MIN: + res = wrapper::vmin(a, b); + break; + case ArithmeticOperation::SQUARED_DIFF: + { + const vec_type tmp = wrapper::vsub(a, b); + res = wrapper::vmul(tmp, tmp); + break; + } + case ArithmeticOperation::PRELU: + { + const vec_type zero = wrapper::vdup_n(static_cast(0), tag_type{}); + const vec_type tmp = wrapper::vmul(a, b); + const auto gt = wrapper::vcgt(a, zero); + + res = wrapper::vbsl(gt, a, tmp); + break; + } + + default: + ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); + } + + return res; +} + +template <> +inline int32x4_t elementwise_arithm_op>(const int32x4_t &a, const int32x4_t &b) +{ + return vcvtq_s32_f32(vfloorq_f32(wrapper::vdiv(vcvtq_f32_s32(a), vcvtq_f32_s32(b)))); +} + +template <> +inline float32x4_t elementwise_arithm_op>(const float32x4_t &a, const float32x4_t &b) +{ + return wrapper::vdiv(a, b); +} + +template <> +inline float32x4_t elementwise_arithm_op>(const float32x4_t &a, const float32x4_t &b) +{ + return wrapper::vpow(a, b); +} + +#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC +template <> +inline float16x8_t elementwise_arithm_op>(const float16x8_t &a, const float16x8_t &b) +{ + return wrapper::vdiv(a, b); +} + +template <> +inline float16x8_t elementwise_arithm_op>(const float16x8_t &a, const float16x8_t &b) +{ + return wrapper::vpow(a, b); +} +#endif // __ARM_FEATURE_FP16_VECTOR_ARITHMETIC + +template +inline typename VectorType::type elementwise_arithm_op_broadcast(const typename VectorType::type &a, const ScalarType &broadcast_value, const bool reorder) +{ + using tag_type = typename VectorType::tag_type; + using vec_type = typename VectorType::type; + + vec_type broadcast_vector = wrapper::vdup_n(broadcast_value, tag_type{}); + return elementwise_arithm_op(reorder ? broadcast_vector : a, reorder ? a : broadcast_vector); +} + +template +inline int elementwise_arithm_op_loop(int window_start_x, int window_end_x, int window_step_x, + const ScalarType *input1_ptr, const ScalarType *input2_ptr, ScalarType *output_ptr) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const auto a = wrapper::vloadq(input1_ptr + x); + const auto b = wrapper::vloadq(input2_ptr + x); + wrapper::vstore(output_ptr + x, elementwise_arithm_op(a, b)); + } + return x; +} + +template +inline int elementwise_arithm_op_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, + const ScalarType *non_broadcast_input_ptr, const ScalarType &broadcast_value, ScalarType *output_ptr, const bool reorder) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const auto a = wrapper::vloadq((non_broadcast_input_ptr + x)); + wrapper::vstore(output_ptr + x, elementwise_arithm_op_broadcast(a, broadcast_value, reorder)); + } + return x; +} + +template +void elementwise_arithm_op(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + using scalar_type = typename VectorType::scalar_type; + + elementwise_op(in1, in2, out, window, + &elementwise_arithm_op_scalar, + &elementwise_arithm_op_broadcast_loop, + &elementwise_arithm_op_loop); +} + +template +inline uint8_t elementwise_comp_op_scalar(const InputScalarType &a, const InputScalarType &b) +{ + bool res = false; + + switch(op) + { + case ComparisonOperation::Equal: + res = (a == b); + break; + case ComparisonOperation::NotEqual: + res = (a != b); + break; + case ComparisonOperation::Greater: + res = (a > b); + break; + case ComparisonOperation::GreaterEqual: + res = (a >= b); + break; + case ComparisonOperation::Less: + res = (a < b); + break; + case ComparisonOperation::LessEqual: + res = (a <= b); + break; + default: + ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); + } + return res ? ~static_cast(0) : static_cast(0); +} + +template +inline OutputVectorType elementwise_comp_op(const InputVectorType &a, const InputVectorType &b) +{ + OutputVectorType res = { 0, 0, 0, 0 }; + + switch(op) + { + case ComparisonOperation::Equal: + res = wrapper::vceq(a, b); + break; + case ComparisonOperation::NotEqual: + res = wrapper::vnot(wrapper::vceq(a, b)); + break; + case ComparisonOperation::Greater: + res = wrapper::vcgt(a, b); + break; + case ComparisonOperation::GreaterEqual: + res = wrapper::vcge(a, b); + break; + case ComparisonOperation::Less: + res = wrapper::vcgt(b, a); + break; + case ComparisonOperation::LessEqual: + res = wrapper::vcge(b, a); + break; + default: + ARM_COMPUTE_ERROR("NOT_SUPPORTED!"); + } + + return res; +} + +template +inline OutputVectorType elementwise_comp_op_broadcast(const InputVectorType &a, const InputScalarType &broadcast_value, const bool reorder) +{ + InputVectorType broadcast_vector = wrapper::vdup_n(broadcast_value, wrapper::traits::vector_128_tag()); + return elementwise_comp_op(reorder ? broadcast_vector : a, reorder ? a : broadcast_vector); +} + +template +inline int elementwise_comp_op_broadcast_8_loop(int window_start_x, int window_end_x, int window_step_x, + const InputScalarType *non_broadcast_input_ptr, const InputScalarType &broadcast_value, uint8_t *output_ptr, const bool reorder) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const auto a = elementwise_comp_op_broadcast(wrapper::vloadq((non_broadcast_input_ptr + x)), broadcast_value, reorder); + wrapper::vstore(output_ptr + x, a); + } + return x; +} + +template +inline int elementwise_comp_op_broadcast_16_loop(int window_start_x, int window_end_x, int window_step_x, + const InputScalarType *non_broadcast_input_ptr, const InputScalarType &broadcast_value, uint8_t *output_ptr, const bool reorder) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const auto a = elementwise_comp_op_broadcast(wrapper::vloadq((non_broadcast_input_ptr + x)), broadcast_value, reorder); + wrapper::vstore(output_ptr + x, wrapper::vmovn(a)); + } + return x; +} + +template +inline int elementwise_comp_op_broadcast_32_loop(int window_start_x, int window_end_x, int window_step_x, + const InputScalarType *non_broadcast_input_ptr, const InputScalarType &broadcast_value, uint8_t *output_ptr, const bool reorder) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const auto a = elementwise_comp_op_broadcast(wrapper::vloadq(non_broadcast_input_ptr + x), broadcast_value, reorder); + const auto b = elementwise_comp_op_broadcast(wrapper::vloadq(non_broadcast_input_ptr + x + 4), broadcast_value, reorder); + wrapper::vstore(output_ptr + x, wrapper::vmovn(wrapper::vcombine(wrapper::vmovn(a), wrapper::vmovn(b)))); + } + if(x <= window_end_x - 4) + { + const auto a = elementwise_comp_op_broadcast(wrapper::vloadq((non_broadcast_input_ptr + x)), broadcast_value, reorder); + for(int i = 0; i < 4; i++) + { + *(output_ptr + x + i) = wrapper::vgetlane(a, i); + } + x = +4; + } + return x; +} + +template +inline int elementwise_comp_op_8_loop(int window_start_x, int window_end_x, int window_step_x, + const InputScalarType *input1_ptr, const InputScalarType *input2_ptr, uint8_t *output_ptr) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const auto a = wrapper::vloadq(input1_ptr + x); + const auto b = wrapper::vloadq(input2_ptr + x); + const auto res = elementwise_comp_op(a, b); + wrapper::vstore(output_ptr + x, res); + } + return x; +} + +template +inline int elementwise_comp_op_16_loop(int window_start_x, int window_end_x, int window_step_x, + const InputScalarType *input1_ptr, const InputScalarType *input2_ptr, uint8_t *output_ptr) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const auto a = wrapper::vloadq(input1_ptr + x); + const auto b = wrapper::vloadq(input2_ptr + x); + const auto res = elementwise_comp_op(a, b); + wrapper::vstore(output_ptr + x, wrapper::vmovn(res)); + } + return x; +} + +template +inline int elementwise_comp_op_32_loop(int window_start_x, int window_end_x, int window_step_x, + const InputScalarType *input1_ptr, const InputScalarType *input2_ptr, uint8_t *output_ptr) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + auto a = wrapper::vloadq(input1_ptr + x); + auto b = wrapper::vloadq(input2_ptr + x); + const auto res = elementwise_comp_op(a, b); + a = wrapper::vloadq(input1_ptr + x + 4); + b = wrapper::vloadq(input2_ptr + x + 4); + const auto res2 = elementwise_comp_op(a, b); + wrapper::vstore(output_ptr + x, wrapper::vmovn(wrapper::vcombine(wrapper::vmovn(res), wrapper::vmovn(res2)))); + } + if(x <= window_end_x - 4) + { + const auto a = wrapper::vloadq(input1_ptr + x); + const auto b = wrapper::vloadq(input2_ptr + x); + const auto res = elementwise_comp_op(a, b); + for(int i = 0; i < 4; i++) + { + *(output_ptr + x + i) = wrapper::vgetlane(res, i); + } + x = +4; + } + return x; +} + +template +void elementwise_comp_op_8(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_op(in1, in2, out, window, + &elementwise_comp_op_scalar, + &elementwise_comp_op_broadcast_8_loop, + &elementwise_comp_op_8_loop); +} + +template +void elementwise_comp_op_16(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_op(in1, in2, out, window, + &elementwise_comp_op_scalar, + &elementwise_comp_op_broadcast_16_loop, + &elementwise_comp_op_16_loop); +} + +template +void elementwise_comp_op_32(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_op(in1, in2, out, window, + &elementwise_comp_op_scalar, + &elementwise_comp_op_broadcast_32_loop, + &elementwise_comp_op_32_loop); +} +} // namesapce cpu +} // namespace arm_compute + +#endif /* SRC_CORE_NEON_KERNELS_ELEMENTWISE_LIST_H */ \ No newline at end of file diff --git a/src/core/NEON/kernels/elementwise/impl/elementwise_quantized_list.h b/src/core/NEON/kernels/elementwise/impl/elementwise_quantized_list.h new file mode 100644 index 0000000000..fd1fb002ad --- /dev/null +++ b/src/core/NEON/kernels/elementwise/impl/elementwise_quantized_list.h @@ -0,0 +1,654 @@ +/* + * Copyright (c) 2021 Arm Limited. + * + * SPDX-License-Identifier: MIT + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in all + * copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE + * SOFTWARE. + */ +#ifndef SRC_CORE_NEON_KERNELS_ELEMENTWISE_QUANTIZED_LIST_H +#define SRC_CORE_NEON_KERNELS_ELEMENTWISE_QUANTIZED_LIST_H + +#include "src/core/NEON/kernels/elementwise/impl/elementwise_list.h" + +namespace arm_compute +{ +namespace cpu +{ +float32x4x4_t load_quantized(const uint8_t *input1_ptr, const int32x4_t &offset, const float32x4_t &scale) +{ + qasymm8x16_t x = vld1q_u8(input1_ptr); + const float32x4x4_t out = + { + { + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_low_u16(vmovl_u8(vget_low_u8(x))))), offset)), scale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_high_u16(vmovl_u8(vget_low_u8(x))))), offset)), scale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_low_u16(vmovl_u8(vget_high_u8(x))))), offset)), scale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vreinterpretq_s32_u32(vmovl_u16(vget_high_u16(vmovl_u8(vget_high_u8(x))))), offset)), scale), + } + }; + return out; +} + +float32x4x4_t load_quantized_signed(const int8_t *input1_ptr, const int32x4_t &offset, const float32x4_t &scale) +{ + qasymm8x16_signed_t x = vld1q_s8(input1_ptr); + const float32x4x4_t out = + { + { + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_low_s16(vmovl_s8(vget_low_s8(x)))), offset)), scale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_high_s16(vmovl_s8(vget_low_s8(x)))), offset)), scale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_low_s16(vmovl_s8(vget_high_s8(x)))), offset)), scale), + vmulq_f32(vcvtq_f32_s32(vsubq_s32(vmovl_s16(vget_high_s16(vmovl_s8(vget_high_s8(x)))), offset)), scale), + } + }; + return out; +} + +void store_quantized(uint8_t *output_ptr, const uint32x4x4_t &out) +{ + const uint8x8_t pa = vqmovn_u16(vcombine_u16(vqmovn_u32(out.val[0]), vqmovn_u32(out.val[1]))); + const uint8x8_t pb = vqmovn_u16(vcombine_u16(vqmovn_u32(out.val[2]), vqmovn_u32(out.val[3]))); + vst1q_u8(output_ptr, vcombine_u8(pa, pb)); +} + +void store_quantized(uint8_t *output_ptr, const int32x4x4_t &out) +{ + const uint8x8_t pa = vqmovun_s16(vcombine_s16(vqmovn_s32(out.val[0]), vqmovn_s32(out.val[1]))); + const uint8x8_t pb = vqmovun_s16(vcombine_s16(vqmovn_s32(out.val[2]), vqmovn_s32(out.val[3]))); + vst1q_u8(output_ptr, vcombine_u8(pa, pb)); +} + +void store_quantized(uint8_t *output_ptr, const float32x4x4_t &rf, const float32x4_t &offset, const float32x4_t &invscale) +{ + int32x4x4_t out = + { + { + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[0], invscale)), + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[1], invscale)), + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[2], invscale)), + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[3], invscale)), + } + }; + store_quantized(output_ptr, out); +} + +void store_quantized_signed(int8_t *output_ptr, const int32x4x4_t &out) +{ + const int8x8_t pa = vqmovn_s16(vcombine_s16(vqmovn_s32(out.val[0]), vqmovn_s32(out.val[1]))); + const int8x8_t pb = vqmovn_s16(vcombine_s16(vqmovn_s32(out.val[2]), vqmovn_s32(out.val[3]))); + vst1q_s8(output_ptr, vcombine_s8(pa, pb)); +} + +void store_quantized_signed(int8_t *output_ptr, const float32x4x4_t &rf, const float32x4_t &offset, const float32x4_t &invscale) +{ + int32x4x4_t out = + { + { + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[0], invscale)), + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[1], invscale)), + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[2], invscale)), + vcvtq_s32_f32(vmlaq_f32(offset, rf.val[3], invscale)), + } + }; + store_quantized_signed(output_ptr, out); +} + +template +inline uint8_t elementwise_arithm_op_quantized_scalar(const float &a, const float &b, UniformQuantizationInfo qinfo) +{ + return quantize_qasymm8(elementwise_arithm_op_scalar(a, b), qinfo); +} + +template +inline int8_t elementwise_arithm_op_quantized_signed_scalar(const float &a, const float &b, UniformQuantizationInfo qinfo) +{ + return quantize_qasymm8_signed(elementwise_arithm_op_scalar(a, b), qinfo); +} + +template +inline float32x4x4_t elementwise_arithm_op(const float32x4x4_t &a, const float32x4x4_t &b) +{ + using neon_vector_float = wrapper::traits::neon_vector; + float32x4x4_t out = + { + { + elementwise_arithm_op(a.val[0], b.val[0]), + elementwise_arithm_op(a.val[1], b.val[1]), + elementwise_arithm_op(a.val[2], b.val[2]), + elementwise_arithm_op(a.val[3], b.val[3]), + } + }; + return out; +} + +template +inline uint8_t elementwise_comp_op_quantized_scalar(const float &a, const float &b, UniformQuantizationInfo qinfo) +{ + ARM_COMPUTE_UNUSED(qinfo); + return elementwise_comp_op_scalar(a, b); +} + +template +inline uint32x4x4_t elementwise_comp_op(const float32x4x4_t &a, const float32x4x4_t &b) +{ + uint32x4x4_t out = + { + { + elementwise_comp_op(a.val[0], b.val[0]), + elementwise_comp_op(a.val[1], b.val[1]), + elementwise_comp_op(a.val[2], b.val[2]), + elementwise_comp_op(a.val[3], b.val[3]) + } + }; + return out; +} + +template +inline int elementwise_arithm_op_quantized_loop(int window_start_x, int window_end_x, int window_step_x, + const uint8_t *input1_ptr, const uint8_t *input2_ptr, uint8_t *output_ptr, + int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, + float32x4_t voffseto, float32x4_t invvscaleo) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + // Get inputs and compute output + const float32x4x4_t af = load_quantized(input1_ptr + x, voffset1, vscale1); + const float32x4x4_t bf = load_quantized(input2_ptr + x, voffset2, vscale2); + const float32x4x4_t rf = elementwise_arithm_op(af, bf); + store_quantized(output_ptr + x, rf, voffseto, invvscaleo); + } + return x; +} + +template +inline int elementwise_arithm_op_quantized_singed_loop(int window_start_x, int window_end_x, int window_step_x, + const int8_t *input1_ptr, const int8_t *input2_ptr, int8_t *output_ptr, + int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, + float32x4_t voffseto, float32x4_t invvscaleo) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + // Get inputs and compute output + const float32x4x4_t af = load_quantized_signed(input1_ptr + x, voffset1, vscale1); + const float32x4x4_t bf = load_quantized_signed(input2_ptr + x, voffset2, vscale2); + const float32x4x4_t rf = elementwise_arithm_op(af, bf); + store_quantized_signed(output_ptr + x, rf, voffseto, invvscaleo); + } + return x; +} + +template +inline int elementwise_arithm_op_quantized_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, + const uint8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, uint8_t *output_ptr, + int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, + float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const float32x4x4_t af = load_quantized(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); + const float32x4x4_t rf = elementwise_arithm_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); + store_quantized(output_ptr + x, rf, voffseto, invvscaleo); + } + return x; +} +template +inline int elementwise_arithm_op_quantized_signed_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, + const int8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, int8_t *output_ptr, + int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, + float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) +{ + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const float32x4x4_t af = load_quantized_signed(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); + const float32x4x4_t rf = elementwise_arithm_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); + store_quantized_signed(output_ptr + x, rf, voffseto, invvscaleo); + } + return x; +} + +template +inline int elementwise_comp_op_quantized_loop(int window_start_x, int window_end_x, int window_step_x, + const uint8_t *input1_ptr, const uint8_t *input2_ptr, uint8_t *output_ptr, + int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, + float32x4_t voffseto, float32x4_t invvscaleo) +{ + ARM_COMPUTE_UNUSED(voffseto, invvscaleo); + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const float32x4x4_t af = load_quantized(input1_ptr + x, voffset1, vscale1); + const float32x4x4_t bf = load_quantized(input2_ptr + x, voffset2, vscale2); + const uint32x4x4_t rf = elementwise_comp_op(af, bf); + store_quantized(output_ptr + x, rf); + } + return x; +} + +template +inline int elementwise_comp_op_quantized_signed_loop(int window_start_x, int window_end_x, int window_step_x, + const int8_t *input1_ptr, const int8_t *input2_ptr, uint8_t *output_ptr, + int32x4_t voffset1, int32x4_t voffset2, float32x4_t vscale1, float32x4_t vscale2, + float32x4_t voffseto, float32x4_t invvscaleo) +{ + ARM_COMPUTE_UNUSED(voffseto, invvscaleo); + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const float32x4x4_t af = load_quantized_signed(input1_ptr + x, voffset1, vscale1); + const float32x4x4_t bf = load_quantized_signed(input2_ptr + x, voffset2, vscale2); + const uint32x4x4_t rf = elementwise_comp_op(af, bf); + store_quantized(output_ptr + x, rf); + } + return x; +} + +template +inline int elementwise_comp_op_quantized_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, + const uint8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, uint8_t *output_ptr, + int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, + float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) +{ + ARM_COMPUTE_UNUSED(voffseto, invvscaleo); + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const float32x4x4_t af = load_quantized(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); + const uint32x4x4_t rf = elementwise_comp_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); + store_quantized(output_ptr + x, rf); + } + return x; +} + +template +inline int elementwise_comp_op_quantized_signed_broadcast_loop(int window_start_x, int window_end_x, int window_step_x, + const int8_t *non_broadcast_input_ptr, float32x4x4_t broadcast_vector, uint8_t *output_ptr, + int32x4_t voffset_non_broadcast, float32x4_t vscale_non_broadcast, + float32x4_t voffseto, float32x4_t invvscaleo, bool reorder) +{ + ARM_COMPUTE_UNUSED(voffseto, invvscaleo); + int x = window_start_x; + for(; x <= (window_end_x - window_step_x); x += window_step_x) + { + const float32x4x4_t af = load_quantized_signed(non_broadcast_input_ptr + x, voffset_non_broadcast, vscale_non_broadcast); + const uint32x4x4_t rf = elementwise_comp_op(reorder ? broadcast_vector : af, reorder ? af : broadcast_vector); + store_quantized(output_ptr + x, rf); + } + return x; +} + +void elementwise_op_quantized(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, + uint8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), + int (*broadcast_func)(int, int, int, const uint8_t *, float32x4x4_t, uint8_t *, int32x4_t, float32x4_t, + float32x4_t, float32x4_t, const bool), + int (*neon_func)(int, int, int, const uint8_t *, const uint8_t *, uint8_t *, + int32x4_t, int32x4_t, float32x4_t, float32x4_t, + float32x4_t, float32x4_t)) +{ + // Create input windows + Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); + Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); + + // Clear X Dimension on execution window as we handle manually + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + const int window_step_x = 16; + const auto window_start_x = static_cast(window.x().start()); + const auto window_end_x = static_cast(window.x().end()); + const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); + + const UniformQuantizationInfo output_qinfo = out->info()->quantization_info().uniform(); + + // Output quantization info (add 0.5 to round toward the nearest integer - 0.5 rounds away from zero) + const float32x4_t voffseto = vdupq_n_f32(output_qinfo.offset + 0.5f); + const float32x4_t invvscaleo = vdupq_n_f32(1.f / output_qinfo.scale); + + if(is_broadcast_across_x) + { + // Select the broadcast input on the X axis + const bool is_broadcast_input_2 = input2_win.x().step() == 0; + Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; + Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; + const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; + const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; + + const UniformQuantizationInfo broadcast_qinfo = broadcast_tensor->info()->quantization_info().uniform(); + const UniformQuantizationInfo non_broadcast_qinfo = non_broadcast_tensor->info()->quantization_info().uniform(); + + const int32x4_t voffset_non_broadcast = vdupq_n_s32(non_broadcast_qinfo.offset); + const float32x4_t vscale_non_broadcast = vdupq_n_f32(non_broadcast_qinfo.scale); + + // Clear X Dimension on execution window as we handle manually + non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator broadcast_input(broadcast_tensor, broadcast_win); + Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); + const auto output_ptr = reinterpret_cast(output.ptr()); + + const uint8_t broadcast_value = *reinterpret_cast(broadcast_input.ptr()); + const float32x4x4_t broadcast_vector = vdequantize(vdupq_n_u8(broadcast_value), broadcast_qinfo); + + int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_vector, output_ptr, + voffset_non_broadcast, vscale_non_broadcast, voffseto, invvscaleo, !is_broadcast_input_2); + for(; x < window_end_x; ++x) + { + const float afs = dequantize_qasymm8(*(non_broadcast_input_ptr + x), non_broadcast_qinfo); + const float bfs = dequantize_qasymm8(broadcast_value, broadcast_qinfo); + *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? bfs : afs, !is_broadcast_input_2 ? afs : bfs, output_qinfo); + } + }, + broadcast_input, non_broadcast_input, output); + } + else + { + const UniformQuantizationInfo input1_qinfo = in1->info()->quantization_info().uniform(); + const UniformQuantizationInfo input2_qinfo = in2->info()->quantization_info().uniform(); + + // Input1 quantization info + const int32x4_t voffset1 = vdupq_n_s32(input1_qinfo.offset); + const float32x4_t vscale1 = vdupq_n_f32(input1_qinfo.scale); + + // Input2 quantization info + const int32x4_t voffset2 = vdupq_n_s32(input2_qinfo.offset); + const float32x4_t vscale2 = vdupq_n_f32(input2_qinfo.scale); + + // Clear X Dimension on execution window as we handle manually + input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator input1(in1, input1_win); + Iterator input2(in2, input2_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + const auto input1_ptr = reinterpret_cast(input1.ptr()); + const auto input2_ptr = reinterpret_cast(input2.ptr()); + const auto output_ptr = reinterpret_cast(output.ptr()); + + int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr, voffset1, voffset2, + vscale1, vscale2, voffseto, invvscaleo); + for(; x < window_end_x; ++x) + { + const float afs = dequantize_qasymm8(*(input1_ptr + x), input1_qinfo); + const float bfs = dequantize_qasymm8(*(input2_ptr + x), input2_qinfo); + *(output_ptr + x) = (*scalar_func)(afs, bfs, output_qinfo); + } + }, + input1, input2, output); + } +} + +void elementwise_comp_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, + uint8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), + int (*broadcast_func)(int, int, int, const int8_t *, float32x4x4_t, uint8_t *, int32x4_t, float32x4_t, + float32x4_t, float32x4_t, const bool), + int (*neon_func)(int, int, int, const int8_t *, const int8_t *, uint8_t *, + int32x4_t, int32x4_t, float32x4_t, float32x4_t, + float32x4_t, float32x4_t)) +{ + // Create input windows + Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); + Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); + + // Clear X Dimension on execution window as we handle manually + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + const int window_step_x = 16; + const auto window_start_x = static_cast(window.x().start()); + const auto window_end_x = static_cast(window.x().end()); + const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); + + const UniformQuantizationInfo output_qinfo = out->info()->quantization_info().uniform(); + + const float32x4_t voffseto = vdupq_n_f32(output_qinfo.offset); + const float32x4_t invvscaleo = vdupq_n_f32(1.f / output_qinfo.scale); + + if(is_broadcast_across_x) + { + // Select the broadcast input on the X axis + const bool is_broadcast_input_2 = input2_win.x().step() == 0; + Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; + Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; + const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; + const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; + + const UniformQuantizationInfo broadcast_qinfo = broadcast_tensor->info()->quantization_info().uniform(); + const UniformQuantizationInfo non_broadcast_qinfo = non_broadcast_tensor->info()->quantization_info().uniform(); + + const int32x4_t voffset_non_broadcast = vdupq_n_s32(non_broadcast_qinfo.offset); + const float32x4_t vscale_non_broadcast = vdupq_n_f32(non_broadcast_qinfo.scale); + + // Clear X Dimension on execution window as we handle manually + non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator broadcast_input(broadcast_tensor, broadcast_win); + Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); + const auto output_ptr = reinterpret_cast(output.ptr()); + + const int8_t broadcast_value = *reinterpret_cast(broadcast_input.ptr()); + const float32x4x4_t broadcast_vector = vdequantize(vdupq_n_s8(broadcast_value), broadcast_qinfo); + + int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_vector, output_ptr, + voffset_non_broadcast, vscale_non_broadcast, voffseto, invvscaleo, !is_broadcast_input_2); + for(; x < window_end_x; ++x) + { + const float afs = dequantize_qasymm8_signed(*(non_broadcast_input_ptr + x), non_broadcast_qinfo); + const float bfs = dequantize_qasymm8_signed(broadcast_value, broadcast_qinfo); + *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? bfs : afs, !is_broadcast_input_2 ? afs : bfs, output_qinfo); + } + }, + broadcast_input, non_broadcast_input, output); + } + else + { + const UniformQuantizationInfo input1_qinfo = in1->info()->quantization_info().uniform(); + const UniformQuantizationInfo input2_qinfo = in2->info()->quantization_info().uniform(); + + // Input1 quantization info + const int32x4_t voffset1 = vdupq_n_s32(input1_qinfo.offset); + const float32x4_t vscale1 = vdupq_n_f32(input1_qinfo.scale); + + // Input2 quantization info + const int32x4_t voffset2 = vdupq_n_s32(input2_qinfo.offset); + const float32x4_t vscale2 = vdupq_n_f32(input2_qinfo.scale); + + // Clear X Dimension on execution window as we handle manually + input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator input1(in1, input1_win); + Iterator input2(in2, input2_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + const auto input1_ptr = reinterpret_cast(input1.ptr()); + const auto input2_ptr = reinterpret_cast(input2.ptr()); + const auto output_ptr = reinterpret_cast(output.ptr()); + + int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr, voffset1, voffset2, + vscale1, vscale2, voffseto, invvscaleo); + for(; x < window_end_x; ++x) + { + const float afs = dequantize_qasymm8_signed(*(input1_ptr + x), input1_qinfo); + const float bfs = dequantize_qasymm8_signed(*(input2_ptr + x), input2_qinfo); + *(output_ptr + x) = (*scalar_func)(afs, bfs, output_qinfo); + } + }, + input1, input2, output); + } +} + +void elementwise_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window, + int8_t (*scalar_func)(const float &, const float &, UniformQuantizationInfo), + int (*broadcast_func)(int, int, int, const int8_t *, float32x4x4_t, int8_t *, int32x4_t, float32x4_t, + float32x4_t, float32x4_t, const bool), + int (*neon_func)(int, int, int, const int8_t *, const int8_t *, int8_t *, + int32x4_t, int32x4_t, float32x4_t, float32x4_t, + float32x4_t, float32x4_t)) +{ + // Create input windows + Window input1_win = window.broadcast_if_dimension_le_one(in1->info()->tensor_shape()); + Window input2_win = window.broadcast_if_dimension_le_one(in2->info()->tensor_shape()); + + // Clear X Dimension on execution window as we handle manually + Window win = window; + win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + const int window_step_x = 16; + const auto window_start_x = static_cast(window.x().start()); + const auto window_end_x = static_cast(window.x().end()); + const bool is_broadcast_across_x = in1->info()->tensor_shape().x() != in2->info()->tensor_shape().x(); + + const UniformQuantizationInfo output_qinfo = out->info()->quantization_info().uniform(); + + const float32x4_t voffseto = vdupq_n_f32(output_qinfo.offset); + const float32x4_t invvscaleo = vdupq_n_f32(1.f / output_qinfo.scale); + + if(is_broadcast_across_x) + { + // Select the broadcast input on the X axis + const bool is_broadcast_input_2 = input2_win.x().step() == 0; + Window broadcast_win = is_broadcast_input_2 ? input2_win : input1_win; + Window non_broadcast_win = !is_broadcast_input_2 ? input2_win : input1_win; + const ITensor *broadcast_tensor = is_broadcast_input_2 ? in2 : in1; + const ITensor *non_broadcast_tensor = !is_broadcast_input_2 ? in2 : in1; + + const UniformQuantizationInfo broadcast_qinfo = broadcast_tensor->info()->quantization_info().uniform(); + const UniformQuantizationInfo non_broadcast_qinfo = non_broadcast_tensor->info()->quantization_info().uniform(); + + const int32x4_t voffset_non_broadcast = vdupq_n_s32(non_broadcast_qinfo.offset); + const float32x4_t vscale_non_broadcast = vdupq_n_f32(non_broadcast_qinfo.scale); + + // Clear X Dimension on execution window as we handle manually + non_broadcast_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator broadcast_input(broadcast_tensor, broadcast_win); + Iterator non_broadcast_input(non_broadcast_tensor, non_broadcast_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + const auto non_broadcast_input_ptr = reinterpret_cast(non_broadcast_input.ptr()); + const auto output_ptr = reinterpret_cast(output.ptr()); + + const int8_t broadcast_value = *reinterpret_cast(broadcast_input.ptr()); + const float32x4x4_t broadcast_vector = vdequantize(vdupq_n_s8(broadcast_value), broadcast_qinfo); + + int x = (*broadcast_func)(window_start_x, window_end_x, window_step_x, non_broadcast_input_ptr, broadcast_vector, output_ptr, + voffset_non_broadcast, vscale_non_broadcast, voffseto, invvscaleo, !is_broadcast_input_2); + for(; x < window_end_x; ++x) + { + const float afs = dequantize_qasymm8_signed(*(non_broadcast_input_ptr + x), non_broadcast_qinfo); + const float bfs = dequantize_qasymm8_signed(broadcast_value, broadcast_qinfo); + *(output_ptr + x) = (*scalar_func)(!is_broadcast_input_2 ? bfs : afs, !is_broadcast_input_2 ? afs : bfs, output_qinfo); + } + }, + broadcast_input, non_broadcast_input, output); + } + else + { + const UniformQuantizationInfo input1_qinfo = in1->info()->quantization_info().uniform(); + const UniformQuantizationInfo input2_qinfo = in2->info()->quantization_info().uniform(); + + // Input1 quantization info + const int32x4_t voffset1 = vdupq_n_s32(input1_qinfo.offset); + const float32x4_t vscale1 = vdupq_n_f32(input1_qinfo.scale); + + // Input2 quantization info + const int32x4_t voffset2 = vdupq_n_s32(input2_qinfo.offset); + const float32x4_t vscale2 = vdupq_n_f32(input2_qinfo.scale); + + // Clear X Dimension on execution window as we handle manually + input1_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + input2_win.set(Window::DimX, Window::Dimension(0, 1, 1)); + + Iterator input1(in1, input1_win); + Iterator input2(in2, input2_win); + Iterator output(out, win); + + execute_window_loop(win, [&](const Coordinates &) + { + const auto input1_ptr = reinterpret_cast(input1.ptr()); + const auto input2_ptr = reinterpret_cast(input2.ptr()); + const auto output_ptr = reinterpret_cast(output.ptr()); + + int x = (*neon_func)(window_start_x, window_end_x, window_step_x, input1_ptr, input2_ptr, output_ptr, voffset1, voffset2, + vscale1, vscale2, voffseto, invvscaleo); + for(; x < window_end_x; ++x) + { + const float afs = dequantize_qasymm8_signed(*(input1_ptr + x), input1_qinfo); + const float bfs = dequantize_qasymm8_signed(*(input2_ptr + x), input2_qinfo); + *(output_ptr + x) = (*scalar_func)(afs, bfs, output_qinfo); + } + }, + input1, input2, output); + } +} + +template +void elementwise_arithm_op_quantized(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_op_quantized(in1, in2, out, window, &elementwise_arithm_op_quantized_scalar, + &elementwise_arithm_op_quantized_broadcast_loop, + &elementwise_arithm_op_quantized_loop); +} +template +void elementwise_arithm_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_op_quantized_signed(in1, in2, out, window, &elementwise_arithm_op_quantized_signed_scalar, + &elementwise_arithm_op_quantized_signed_broadcast_loop, + &elementwise_arithm_op_quantized_singed_loop); +} + +template +void elementwise_comp_op_quantized(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_op_quantized(in1, in2, out, window, &elementwise_comp_op_quantized_scalar, + &elementwise_comp_op_quantized_broadcast_loop, + &elementwise_comp_op_quantized_loop); +} + +template +void elementwise_comp_op_quantized_signed(const ITensor *in1, const ITensor *in2, ITensor *out, const Window &window) +{ + elementwise_comp_quantized_signed(in1, in2, out, window, &elementwise_comp_op_quantized_scalar, + &elementwise_comp_op_quantized_signed_broadcast_loop, + &elementwise_comp_op_quantized_signed_loop); +} +} // namespace cpu +} // namespace arm_compute + +#endif /* SRC_CORE_NEON_KERNELS_ELEMENTWISE_QUANTIZED_LIST_H */ -- cgit v1.2.1