COMPMID-1566: Add broadcast to CLArithmeticSubtraction

Change-Id: I05d21f9a92013ecfd1128d12cf1561cfd6e5c5e9
Reviewed-on: https://eu-gerrit-1.euhpc.arm.com/147983
Tested-by: bsgcomp <bsgcomp@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>

4 files changed, 77 insertions, 148 deletions

diff --git a/tests/validation/fixtures/ArithmeticAdditionFixture.h b/tests/validation/fixtures/ArithmeticOperationsFixture.h
index 66875934de..fb46a5185e 100644
--- a/tests/validation/fixtures/ArithmeticAdditionFixture.h
+++ b/tests/validation/fixtures/ArithmeticOperationsFixture.h
@@ -21,8 +21,8 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef ARM_COMPUTE_TEST_ARITHMETIC_ADDITION_FIXTURE
-#define ARM_COMPUTE_TEST_ARITHMETIC_ADDITION_FIXTURE
+#ifndef ARM_COMPUTE_TEST_ARITHMETIC_OPERATIONS_FIXTURE
+#define ARM_COMPUTE_TEST_ARITHMETIC_OPERATIONS_FIXTURE
 
 #include "arm_compute/core/TensorShape.h"
 #include "arm_compute/core/Types.h"
@@ -32,7 +32,7 @@
 #include "tests/framework/Asserts.h"
 #include "tests/framework/Fixture.h"
 #include "tests/validation/Helpers.h"
-#include "tests/validation/reference/ArithmeticAddition.h"
+#include "tests/validation/reference/ArithmeticOperations.h"
 
 namespace arm_compute
 {
@@ -41,13 +41,15 @@ namespace test
 namespace validation
 {
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
-class ArithmeticAdditionGenericFixture : public framework::Fixture
+class ArithmeticOperationGenericFixture : public framework::Fixture
 {
 public:
     template <typename...>
-    void setup(const TensorShape &shape0, const TensorShape &shape1, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy,
+    void setup(reference::ArithmeticOperation op, const TensorShape &shape0, const TensorShape &shape1,
+               DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy,
                QuantizationInfo qinfo0, QuantizationInfo qinfo1, QuantizationInfo qinfo_out)
     {
+        _op        = op;
         _target    = compute_target(shape0, shape1, data_type0, data_type1, output_data_type, convert_policy, qinfo0, qinfo1, qinfo_out);
         _reference = compute_reference(shape0, shape1, data_type0, data_type1, output_data_type, convert_policy, qinfo0, qinfo1, qinfo_out);
     }
@@ -68,8 +70,8 @@ protected:
         TensorType dst      = create_tensor<TensorType>(TensorShape::broadcast_shape(shape0, shape1), output_data_type, 1, qinfo_out);
 
         // Create and configure function
-        FunctionType add;
-        add.configure(&ref_src1, &ref_src2, &dst, convert_policy);
+        FunctionType arith_op;
+        arith_op.configure(&ref_src1, &ref_src2, &dst, convert_policy);
 
         ARM_COMPUTE_EXPECT(ref_src1.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(ref_src2.info()->is_resizable(), framework::LogLevel::ERRORS);
@@ -89,7 +91,7 @@ protected:
         fill(AccessorType(ref_src2), 1);
 
         // Compute function
-        add.run();
+        arith_op.run();
 
         return dst;
     }
@@ -107,39 +109,40 @@ protected:
         fill(ref_src1, 0);
         fill(ref_src2, 1);
 
-        return reference::arithmetic_addition<T>(ref_src1, ref_src2, ref_dst, convert_policy);
+        return reference::arithmetic_operation<T>(_op, ref_src1, ref_src2, ref_dst, convert_policy);
     }
 
-    TensorType      _target{};
-    SimpleTensor<T> _reference{};
+    TensorType                     _target{};
+    SimpleTensor<T>                _reference{};
+    reference::ArithmeticOperation _op{ reference::ArithmeticOperation::ADD };
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
-class ArithmeticAdditionBroadcastValidationFixture : public ArithmeticAdditionGenericFixture<TensorType, AccessorType, FunctionType, T>
+class ArithmeticAdditionBroadcastValidationFixture : public ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>
 {
 public:
     template <typename...>
     void setup(const TensorShape &shape0, const TensorShape &shape1, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
     {
-        ArithmeticAdditionGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(shape0, shape1, data_type0, data_type1,
-                                                                                           output_data_type, convert_policy, QuantizationInfo(), QuantizationInfo(), QuantizationInfo());
+        ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(reference::ArithmeticOperation::ADD, shape0, shape1, data_type0, data_type1,
+                                                                                            output_data_type, convert_policy, QuantizationInfo(), QuantizationInfo(), QuantizationInfo());
     }
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
-class ArithmeticAdditionValidationFixture : public ArithmeticAdditionGenericFixture<TensorType, AccessorType, FunctionType, T>
+class ArithmeticAdditionValidationFixture : public ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>
 {
 public:
     template <typename...>
     void setup(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
     {
-        ArithmeticAdditionGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(shape, shape, data_type0, data_type1,
-                                                                                           output_data_type, convert_policy, QuantizationInfo(), QuantizationInfo(), QuantizationInfo());
+        ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(reference::ArithmeticOperation::ADD, shape, shape, data_type0, data_type1,
+                                                                                            output_data_type, convert_policy, QuantizationInfo(), QuantizationInfo(), QuantizationInfo());
     }
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
-class ArithmeticAdditionValidationQuantizedFixture : public ArithmeticAdditionGenericFixture<TensorType, AccessorType, FunctionType, T>
+class ArithmeticAdditionValidationQuantizedFixture : public ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>
 {
 public:
     template <typename...>
@@ -147,11 +150,52 @@ public:
                QuantizationInfo qinfo0, QuantizationInfo qinfo1, QuantizationInfo qinfo_out)
 
     {
-        ArithmeticAdditionGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(shape, shape, data_type0, data_type1,
-                                                                                           output_data_type, convert_policy, qinfo0, qinfo1, qinfo_out);
+        ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(reference::ArithmeticOperation::ADD, shape, shape, data_type0, data_type1,
+                                                                                            output_data_type, convert_policy, qinfo0, qinfo1, qinfo_out);
+    }
+};
+
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ArithmeticSubtractionBroadcastValidationFixture : public ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>
+{
+public:
+    template <typename...>
+    void setup(const TensorShape &shape0, const TensorShape &shape1, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
+    {
+        ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(reference::ArithmeticOperation::SUB, shape0, shape1,
+                                                                                            data_type0, data_type1, output_data_type, convert_policy,
+                                                                                            QuantizationInfo(), QuantizationInfo(), QuantizationInfo());
+    }
+};
+
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ArithmeticSubtractionValidationFixture : public ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>
+{
+public:
+    template <typename...>
+    void setup(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
+    {
+        ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(reference::ArithmeticOperation::SUB, shape, shape,
+                                                                                            data_type0, data_type1, output_data_type, convert_policy,
+                                                                                            QuantizationInfo(), QuantizationInfo(), QuantizationInfo());
+    }
+};
+
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ArithmeticSubtractionValidationQuantizedFixture : public ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>
+{
+public:
+    template <typename...>
+    void setup(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy,
+               QuantizationInfo qinfo0, QuantizationInfo qinfo1, QuantizationInfo qinfo_out)
+
+    {
+        ArithmeticOperationGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(reference::ArithmeticOperation::SUB, shape, shape,
+                                                                                            data_type0, data_type1, output_data_type,
+                                                                                            convert_policy, qinfo0, qinfo1, qinfo_out);
     }
 };
 } // namespace validation
 } // namespace test
 } // namespace arm_compute
-#endif /* ARM_COMPUTE_TEST_ARITHMETIC_ADDITION_FIXTURE */
+#endif /* ARM_COMPUTE_TEST_ARITHMETIC_OPERATIONS_FIXTURE */
diff --git a/tests/validation/fixtures/ArithmeticSubtractionFixture.h b/tests/validation/fixtures/ArithmeticSubtractionFixture.h
deleted file mode 100644
index 9897677691..0000000000
--- a/tests/validation/fixtures/ArithmeticSubtractionFixture.h
+++ /dev/null
@@ -1,114 +0,0 @@
-/*
- * Copyright (c) 2017-2018 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 ARM_COMPUTE_TEST_ARITHMETIC_SUBTRACTION_FIXTURE
-#define ARM_COMPUTE_TEST_ARITHMETIC_SUBTRACTION_FIXTURE
-
-#include "arm_compute/core/TensorShape.h"
-#include "arm_compute/core/Types.h"
-#include "tests/AssetsLibrary.h"
-#include "tests/Globals.h"
-#include "tests/IAccessor.h"
-#include "tests/framework/Asserts.h"
-#include "tests/framework/Fixture.h"
-#include "tests/validation/Helpers.h"
-#include "tests/validation/reference/ArithmeticSubtraction.h"
-
-namespace arm_compute
-{
-namespace test
-{
-namespace validation
-{
-template <typename TensorType, typename AccessorType, typename FunctionType, typename T1, typename T2 = T1, typename T3 = T1>
-class ArithmeticSubtractionValidationFixture : public framework::Fixture
-{
-public:
-    template <typename...>
-    void setup(TensorShape shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
-    {
-        _target    = compute_target(shape, data_type0, data_type1, output_data_type, convert_policy);
-        _reference = compute_reference(shape, data_type0, data_type1, output_data_type, convert_policy);
-    }
-
-protected:
-    template <typename U>
-    void fill(U &&tensor, int i)
-    {
-        library->fill_tensor_uniform(tensor, i);
-    }
-
-    TensorType compute_target(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
-    {
-        // Create tensors
-        TensorType ref_src1 = create_tensor<TensorType>(shape, data_type0, 1);
-        TensorType ref_src2 = create_tensor<TensorType>(shape, data_type1, 1);
-        TensorType dst      = create_tensor<TensorType>(shape, output_data_type, 1);
-
-        // Create and configure function
-        FunctionType sub;
-        sub.configure(&ref_src1, &ref_src2, &dst, convert_policy);
-
-        ARM_COMPUTE_EXPECT(ref_src1.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(ref_src2.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
-
-        // Allocate tensors
-        ref_src1.allocator()->allocate();
-        ref_src2.allocator()->allocate();
-        dst.allocator()->allocate();
-
-        ARM_COMPUTE_EXPECT(!ref_src1.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(!ref_src2.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(!dst.info()->is_resizable(), framework::LogLevel::ERRORS);
-
-        // Fill tensors
-        fill(AccessorType(ref_src1), 0);
-        fill(AccessorType(ref_src2), 1);
-
-        // Compute function
-        sub.run();
-
-        return dst;
-    }
-
-    SimpleTensor<T3> compute_reference(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
-    {
-        // Create reference
-        SimpleTensor<T1> ref_src1{ shape, data_type0, 1 };
-        SimpleTensor<T2> ref_src2{ shape, data_type1, 1 };
-
-        // Fill reference
-        fill(ref_src1, 0);
-        fill(ref_src2, 1);
-
-        return reference::arithmetic_subtraction<T1, T2, T3>(ref_src1, ref_src2, output_data_type, convert_policy);
-    }
-
-    TensorType       _target{};
-    SimpleTensor<T3> _reference{};
-};
-} // namespace validation
-} // namespace test
-} // namespace arm_compute
-#endif /* ARM_COMPUTE_TEST_ARITHMETIC_SUBTRACTION_FIXTURE */
diff --git a/tests/validation/fixtures/LSTMLayerFixture.h b/tests/validation/fixtures/LSTMLayerFixture.h
index 20df855242..bc892bfecf 100644
--- a/tests/validation/fixtures/LSTMLayerFixture.h
+++ b/tests/validation/fixtures/LSTMLayerFixture.h
@@ -28,8 +28,7 @@
 #include "tests/framework/Asserts.h"
 #include "tests/framework/Fixture.h"
 #include "tests/validation/reference/ActivationLayer.h"
-#include "tests/validation/reference/ArithmeticAddition.h"
-#include "tests/validation/reference/ArithmeticSubtraction.h"
+#include "tests/validation/reference/ArithmeticOperations.h"
 #include "tests/validation/reference/FullyConnectedLayer.h"
 #include "tests/validation/reference/GEMM.h"
 #include "tests/validation/reference/PixelWiseMultiplication.h"
@@ -333,12 +332,12 @@ protected:
         SimpleTensor<T> fully_connected_forget = reference::fully_connected_layer(input, input_to_forget_w, forget_gate_bias, output_cell_shape);
         SimpleTensor<T> transposed_weights     = reference::transpose(recurrent_to_forget_w);
         SimpleTensor<T> gemm                   = reference::gemm(output_state_in, transposed_weights, cell_state_in, 1.f, 0.f);
-        SimpleTensor<T> forget_gate            = reference::arithmetic_addition(fully_connected_forget, gemm, data_type, ConvertPolicy::SATURATE);
+        SimpleTensor<T> forget_gate            = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, fully_connected_forget, gemm, data_type, ConvertPolicy::SATURATE);
 
         if(peephole_opt)
         {
             SimpleTensor<T> pixelwise_mul_forget_gate = reference::pixel_wise_multiplication(cell_state_in, cell_to_forget_w, 1, ConvertPolicy::SATURATE, RoundingPolicy::TO_NEAREST_EVEN);
-            forget_gate                               = reference::arithmetic_addition(forget_gate, pixelwise_mul_forget_gate, data_type, ConvertPolicy::SATURATE);
+            forget_gate                               = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, forget_gate, pixelwise_mul_forget_gate, data_type, ConvertPolicy::SATURATE);
         }
 
         forget_gate = reference::activation_layer(forget_gate, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LOGISTIC));
@@ -349,18 +348,18 @@ protected:
         {
             SimpleTensor<T> ones{ cell_bias_shape, data_type };
             fill_custom_val(ones, 1.f, 0);
-            input_gate = reference::arithmetic_subtraction<T, T, T>(ones, forget_gate, data_type, ConvertPolicy::SATURATE);
+            input_gate = reference::arithmetic_operation<T>(reference::ArithmeticOperation::SUB, ones, forget_gate, data_type, ConvertPolicy::SATURATE);
         }
         else
         {
             SimpleTensor<T> fully_connected_input = reference::fully_connected_layer(input, input_to_input_w, input_gate_bias, output_cell_shape);
             transposed_weights                    = reference::transpose(recurrent_to_input_w);
             gemm                                  = reference::gemm(output_state_in, transposed_weights, cell_state_in, 1.f, 0.f);
-            input_gate                            = reference::arithmetic_addition(fully_connected_input, gemm, data_type, ConvertPolicy::SATURATE);
+            input_gate                            = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, fully_connected_input, gemm, data_type, ConvertPolicy::SATURATE);
             if(peephole_opt)
             {
                 SimpleTensor<T> pixelwise_mul_input_gate = reference::pixel_wise_multiplication(cell_state_in, cell_to_input_w, 1, ConvertPolicy::SATURATE, RoundingPolicy::TO_NEAREST_EVEN);
-                input_gate                               = reference::arithmetic_addition(input_gate, pixelwise_mul_input_gate, data_type, ConvertPolicy::SATURATE);
+                input_gate                               = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, input_gate, pixelwise_mul_input_gate, data_type, ConvertPolicy::SATURATE);
             }
             input_gate = reference::activation_layer(input_gate, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LOGISTIC));
         }
@@ -370,10 +369,10 @@ protected:
         transposed_weights                         = reference::transpose(recurrent_to_cell_w);
         gemm                                       = reference::gemm(output_state_in, transposed_weights, cell_state_out, 1.f, 0.f);
         SimpleTensor<T> pixelwise_mul              = reference::pixel_wise_multiplication(cell_state_in, forget_gate, 1, ConvertPolicy::SATURATE, RoundingPolicy::TO_NEAREST_EVEN);
-        cell_state_out                             = reference::arithmetic_addition(fully_connected_cell_state, gemm, data_type, ConvertPolicy::SATURATE);
+        cell_state_out                             = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, fully_connected_cell_state, gemm, data_type, ConvertPolicy::SATURATE);
         cell_state_out                             = reference::activation_layer(cell_state_out, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LOGISTIC));
         cell_state_out                             = reference::pixel_wise_multiplication(cell_state_out, input_gate, 1, ConvertPolicy::SATURATE, RoundingPolicy::TO_NEAREST_EVEN);
-        cell_state_out                             = reference::arithmetic_addition(cell_state_out, pixelwise_mul, data_type, ConvertPolicy::SATURATE);
+        cell_state_out                             = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, cell_state_out, pixelwise_mul, data_type, ConvertPolicy::SATURATE);
         if(cell_threshold != 0.f)
         {
             cell_state_out = reference::activation_layer(cell_state_out, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LU_BOUNDED_RELU, -cell_threshold, cell_threshold));
@@ -383,11 +382,11 @@ protected:
         SimpleTensor<T> fully_connected_output = reference::fully_connected_layer(input, input_to_output_w, output_gate_bias, output_cell_shape);
         transposed_weights                     = reference::transpose(recurrent_to_output_w);
         gemm                                   = reference::gemm(output_state_in, transposed_weights, cell_state_out, 1.f, 0.f);
-        output                                 = reference::arithmetic_addition(fully_connected_output, gemm, data_type, ConvertPolicy::SATURATE);
+        output                                 = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, fully_connected_output, gemm, data_type, ConvertPolicy::SATURATE);
         if(peephole_opt)
         {
             pixelwise_mul = reference::pixel_wise_multiplication(cell_state_out, cell_to_output_w, 1, ConvertPolicy::SATURATE, RoundingPolicy::TO_NEAREST_EVEN);
-            output        = reference::arithmetic_addition(output, pixelwise_mul, data_type, ConvertPolicy::SATURATE);
+            output        = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, output, pixelwise_mul, data_type, ConvertPolicy::SATURATE);
         }
         output = reference::activation_layer(output, ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::LOGISTIC));
 
diff --git a/tests/validation/fixtures/RNNLayerFixture.h b/tests/validation/fixtures/RNNLayerFixture.h
index 42b99cce1c..2645116b44 100644
--- a/tests/validation/fixtures/RNNLayerFixture.h
+++ b/tests/validation/fixtures/RNNLayerFixture.h
@@ -28,7 +28,7 @@
 #include "tests/framework/Asserts.h"
 #include "tests/framework/Fixture.h"
 #include "tests/validation/reference/ActivationLayer.h"
-#include "tests/validation/reference/ArithmeticAddition.h"
+#include "tests/validation/reference/ArithmeticOperations.h"
 #include "tests/validation/reference/FullyConnectedLayer.h"
 #include "tests/validation/reference/GEMM.h"
 
@@ -132,7 +132,7 @@ protected:
         SimpleTensor<T> out_w{ out_shape, data_type };
         SimpleTensor<T> fully_connected = reference::fully_connected_layer(input, weights, bias, out_shape);
         SimpleTensor<T> gemm            = reference::gemm(hidden_state, recurrent_weights, out_w, 1.f, 0.f);
-        SimpleTensor<T> add_res         = reference::arithmetic_addition(fully_connected, gemm, data_type, ConvertPolicy::SATURATE);
+        SimpleTensor<T> add_res         = reference::arithmetic_operation(reference::ArithmeticOperation::ADD, fully_connected, gemm, data_type, ConvertPolicy::SATURATE);
         return reference::activation_layer(add_res, info);
     }