COMPMID-1717: CL: Implement Maximum, Minimum, SquaredDifference

Change-Id: Ice653e48211053bd3cd20a693bd76de6b4efc370
Reviewed-on: https://review.mlplatform.org/270
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

5 files changed, 195 insertions, 252 deletions

diff --git a/arm_compute/core/CL/CLKernels.h b/arm_compute/core/CL/CLKernels.h
index c7c12975e0..c707265c23 100644
--- a/arm_compute/core/CL/CLKernels.h
+++ b/arm_compute/core/CL/CLKernels.h
@@ -28,9 +28,6 @@
 #include "arm_compute/core/CL/kernels/CLAbsoluteDifferenceKernel.h"
 #include "arm_compute/core/CL/kernels/CLAccumulateKernel.h"
 #include "arm_compute/core/CL/kernels/CLActivationLayerKernel.h"
-#include "arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h"
-#include "arm_compute/core/CL/kernels/CLArithmeticDivisionKernel.h"
-#include "arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h"
 #include "arm_compute/core/CL/kernels/CLBatchNormalizationLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLBatchToSpaceLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLBitwiseAndKernel.h"
@@ -62,6 +59,7 @@
 #include "arm_compute/core/CL/kernels/CLDilateKernel.h"
 #include "arm_compute/core/CL/kernels/CLDirectConvolutionLayerKernel.h"
 #include "arm_compute/core/CL/kernels/CLDirectConvolutionLayerOutputStageKernel.h"
+#include "arm_compute/core/CL/kernels/CLElementwiseOperationKernel.h"
 #include "arm_compute/core/CL/kernels/CLErodeKernel.h"
 #include "arm_compute/core/CL/kernels/CLFastCornersKernel.h"
 #include "arm_compute/core/CL/kernels/CLFillBorderKernel.h"
diff --git a/arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h b/arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h
deleted file mode 100644
index 48e72f3c13..0000000000
--- a/arm_compute/core/CL/kernels/CLArithmeticAdditionKernel.h
+++ /dev/null
@@ -1,83 +0,0 @@
-/*
- * Copyright (c) 2016-2018 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_CLARITHMETICADDITIONKERNEL_H__
-#define __ARM_COMPUTE_CLARITHMETICADDITIONKERNEL_H__
-
-#include "arm_compute/core/CL/ICLKernel.h"
-#include "arm_compute/core/Types.h"
-
-namespace arm_compute
-{
-class ICLTensor;
-
-/** Interface for the arithmetic addition kernel
- *
- * Arithmetic addition is computed by:
- * @f[ output(x,y) = input1(x,y) + input2(x,y) @f]
- */
-class CLArithmeticAdditionKernel : public ICLKernel
-{
-public:
-    /** Default constructor */
-    CLArithmeticAdditionKernel();
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    CLArithmeticAdditionKernel(const CLArithmeticAdditionKernel &) = delete;
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    CLArithmeticAdditionKernel &operator=(const CLArithmeticAdditionKernel &) = delete;
-    /** Allow instances of this class to be moved */
-    CLArithmeticAdditionKernel(CLArithmeticAdditionKernel &&) = default;
-    /** Allow instances of this class to be moved */
-    CLArithmeticAdditionKernel &operator=(CLArithmeticAdditionKernel &&) = default;
-    /** Default destructor */
-    ~CLArithmeticAdditionKernel() = default;
-    /** Initialise the kernel's inputs, output and conversion policy.
-     *
-     * @param[in]  input1 First tensor input. Data types supported: U8/QASYMM8/S16/F16/F32.
-     * @param[in]  input2 Second tensor input. Data types supported: U8, QASYMM8 (only if @p input1 is QASYMM8), S16/F16/F32.
-     * @param[out] output Output tensor. Data types supported: U8 (Only if both inputs are U8), QASYMM8 (only if @p input1 is QASYMM8), S16/F16/F32.
-     * @param[in]  policy Policy to use to handle overflow.
-     */
-    void configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticAdditionKernel
-     *
-     * @param[in] input1 First tensor input info. Data types supported: U8/QASYMM8/S16/F16/F32.
-     * @param[in] input2 Second tensor input info. Data types supported: U8, QASYMM8 (only if @p input1 is QASYMM8), S16/F16/F32.
-     * @param[in] output Output tensor info. Data types supported: U8 (Only if both inputs are U8), QASYMM8 (only if both inputs are QASYMM8), S16/F16/F32.
-     * @param[in] policy Policy to use to handle overflow.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy);
-
-    // Inherited methods overridden:
-    void run(const Window &window, cl::CommandQueue &queue) override;
-    BorderSize border_size() const override;
-
-private:
-    const ICLTensor *_input1; /**< Source tensor 1 */
-    const ICLTensor *_input2; /**< Source tensor 2 */
-    ICLTensor       *_output; /**< Destination tensor */
-};
-} // namespace arm_compute
-#endif /* __ARM_COMPUTE_CLARITHMETICADDITIONKERNEL_H__ */
diff --git a/arm_compute/core/CL/kernels/CLArithmeticDivisionKernel.h b/arm_compute/core/CL/kernels/CLArithmeticDivisionKernel.h
deleted file mode 100644
index 430a641559..0000000000
--- a/arm_compute/core/CL/kernels/CLArithmeticDivisionKernel.h
+++ /dev/null
@@ -1,81 +0,0 @@
-/*
- * Copyright (c) 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_CLARITHMETICDIVISIONKERNEL_H__
-#define __ARM_COMPUTE_CLARITHMETICDIVISIONKERNEL_H__
-
-#include "arm_compute/core/CL/ICLKernel.h"
-#include "arm_compute/core/Types.h"
-
-namespace arm_compute
-{
-class ICLTensor;
-
-/** Interface for the arithmetic division kernel
- *
- * Arithmetic division is computed by:
- * @f[ output(x,y) = input1(x,y) / input2(x,y) @f]
- */
-class CLArithmeticDivisionKernel : public ICLKernel
-{
-public:
-    /** Default constructor */
-    CLArithmeticDivisionKernel();
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    CLArithmeticDivisionKernel(const CLArithmeticDivisionKernel &) = delete;
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    CLArithmeticDivisionKernel &operator=(const CLArithmeticDivisionKernel &) = delete;
-    /** Allow instances of this class to be moved */
-    CLArithmeticDivisionKernel(CLArithmeticDivisionKernel &&) = default;
-    /** Allow instances of this class to be moved */
-    CLArithmeticDivisionKernel &operator=(CLArithmeticDivisionKernel &&) = default;
-    /** Default destructor */
-    ~CLArithmeticDivisionKernel() = default;
-    /** Initialise the kernel's inputs, output.
-     *
-     * @param[in]  input1 First tensor input. Data types supported: F16/F32.
-     * @param[in]  input2 Second tensor input. Data types supported: Same as @p input1.
-     * @param[out] output Output tensor. Data types supported: Same as @p input1.
-     */
-    void configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticDivisionKernel
-     *
-     * @param[in] input1 First tensor input info. Data types supported: F16/F32.
-     * @param[in] input2 Second tensor input info. Data types supported: Same as @p input1.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input1.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output);
-
-    // Inherited methods overridden:
-    void run(const Window &window, cl::CommandQueue &queue) override;
-    BorderSize border_size() const override;
-
-private:
-    const ICLTensor *_input1; /**< Source tensor 1 */
-    const ICLTensor *_input2; /**< Source tensor 2 */
-    ICLTensor       *_output; /**< Destination tensor */
-};
-} // namespace arm_compute
-#endif /* __ARM_COMPUTE_CLARITHMETICDIVISIONKERNEL_H__ */
diff --git a/arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h b/arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h
deleted file mode 100644
index 9875ac7a31..0000000000
--- a/arm_compute/core/CL/kernels/CLArithmeticSubtractionKernel.h
+++ /dev/null
@@ -1,85 +0,0 @@
-/*
- * Copyright (c) 2016-2018 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_CLARITHMETICSUBTRACTIONKERNEL_H__
-#define __ARM_COMPUTE_CLARITHMETICSUBTRACTIONKERNEL_H__
-
-#include "arm_compute/core/CL/ICLKernel.h"
-
-#include "arm_compute/core/Types.h"
-
-namespace arm_compute
-{
-class ICLTensor;
-
-/** Interface for the arithmetic subtraction kernel
- *
- * Arithmetic subtraction is computed by:
- * @f[ output(x,y) = input1(x,y) - input2(x,y) @f]
- */
-class CLArithmeticSubtractionKernel : public ICLKernel
-{
-public:
-    /** Default constructor */
-    CLArithmeticSubtractionKernel();
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    CLArithmeticSubtractionKernel(const CLArithmeticSubtractionKernel &) = delete;
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    CLArithmeticSubtractionKernel &operator=(const CLArithmeticSubtractionKernel &) = delete;
-    /** Allow instances of this class to be moved */
-    CLArithmeticSubtractionKernel(CLArithmeticSubtractionKernel &&) = default;
-    /** Allow instances of this class to be moved */
-    CLArithmeticSubtractionKernel &operator=(CLArithmeticSubtractionKernel &&) = default;
-    /** Default destructor */
-    ~CLArithmeticSubtractionKernel() = default;
-
-    /** Initialise the kernel's inputs, output and conversion policy.
-     *
-     * @param[in]  input1 First tensor input. Data types supported: U8/QASYMM8/S16/F16/F32.
-     * @param[in]  input2 Second tensor input. Data types supported: U8/QASYMM8/S16/F16/F32.
-     * @param[out] output Output tensor. Data types supported: U8 (Only if both inputs are U8), QASYMM8/S16/F16/F32.
-     * @param[in]  policy Policy to use to handle overflow.
-     */
-    void configure(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, ConvertPolicy policy);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticSubtractionKernel
-     *
-     * @param[in] input1 First tensor input info. Data types supported: U8/QASYMM8/S16/F16/F32.
-     * @param[in] input2 Second tensor input info. Data types supported: U8/QASYMM8/S16/F16/F32.
-     * @param[in] output Output tensor info. Data types supported: U8 (Only if both inputs are U8), QASYMM8/S16/F16/F32.
-     * @param[in] policy Policy to use to handle overflow.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, ConvertPolicy policy);
-
-    // Inherited methods overridden:
-    void run(const Window &window, cl::CommandQueue &queue) override;
-    BorderSize border_size() const override;
-
-private:
-    const ICLTensor *_input1; /**< Source tensor 1 */
-    const ICLTensor *_input2; /**< Source tensor 2 */
-    ICLTensor       *_output; /**< Destination tensor */
-};
-} // namespace arm_compute
-#endif /* __ARM_COMPUTE_CLARITHMETICSUBTRACTIONKERNEL_H__ */
diff --git a/arm_compute/core/CL/kernels/CLElementwiseOperationKernel.h b/arm_compute/core/CL/kernels/CLElementwiseOperationKernel.h
new file mode 100644
index 0000000000..2c65789115
--- /dev/null
+++ b/arm_compute/core/CL/kernels/CLElementwiseOperationKernel.h
@@ -0,0 +1,194 @@
+/*
+ * Copyright (c) 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_CLELEMENTWISEOPERATIONKERNEL_H__
+#define __ARM_COMPUTE_CLELEMENTWISEOPERATIONKERNEL_H__
+
+#include "arm_compute/core/CL/ICLKernel.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+class ICLTensor;
+
+/** Interface for an element-wise operation kernel
+ *
+ * Element-wise operation is computed by:
+ * @f[ output(x,y) = OP(input1(x,y), input2(x,y))@f]
+ *
+ */
+class CLElementwiseOperationKernel : public ICLKernel
+{
+public:
+    /** Default constructor */
+    CLElementwiseOperationKernel();
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    CLElementwiseOperationKernel(const CLElementwiseOperationKernel &) = delete;
+    /** Prevent instances of this class from being copied (As this class contains pointers) */
+    CLElementwiseOperationKernel &operator=(const CLElementwiseOperationKernel &) = delete;
+    /** Allow instances of this class to be moved */
+    CLElementwiseOperationKernel(CLElementwiseOperationKernel &&) = default;
+    /** Allow instances of this class to be moved */
+    CLElementwiseOperationKernel &operator=(CLElementwiseOperationKernel &&) = default;
+    /** Default destructor */
+    ~CLElementwiseOperationKernel() = default;
+
+    // Inherited methods overridden:
+    void run(const Window &window, cl::CommandQueue &queue) override;
+
+    BorderSize border_size() const override;
+
+protected:
+    /** The name of the operation */
+    virtual std::string name() = 0;
+
+    /** Initialise the kernel's output.
+     *
+     * @param[in] input1 First tensor input. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input2 Second tensor input. Data types supported: Same as @p input1.
+     * @param[in] output Output tensor. Data types supported: Same as @p input1.
+     *
+     * @return a pair of Status and Window
+     */
+    virtual std::pair<Status, Window> validate_and_configure_window(ITensorInfo &input1, ITensorInfo &input2, ITensorInfo &output) = 0;
+
+    /** Validate the argument passed to the kernel
+     *
+     * @param[in] input1 First tensor input. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input2 Second tensor input. Data types supported: Same as @p input1.
+     * @param[in] output Output tensor. Data types supported: Same as @p input1.
+     */
+    virtual Status validate_arguments(const ITensorInfo &input1, const ITensorInfo &input2, const ITensorInfo &output) = 0;
+
+    /** Generate the build options for the specific kernel
+     *
+     * @reutrn a CLBuildOptions struct
+     */
+    virtual CLBuildOptions generate_build_options(const ITensorInfo &input1, const ITensorInfo &input2, const ITensorInfo &output) = 0;
+
+    /** Generate the identifier for tuning
+     *
+     * @reutrn a string
+     */
+    virtual std::string generate_id_for_tuning(const std::string &kernel_name, const ITensorInfo &input1, const ITensorInfo &output) = 0;
+
+    /** Commmon configure function for element-wise operators with no additional options (e.g., Div, Min, Max, SquaredDiff)
+     *
+     */
+    void configure_common(const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output);
+
+private:
+    const ICLTensor *_input1; /**< Source tensor 1 */
+    const ICLTensor *_input2; /**< Source tensor 2 */
+    ICLTensor       *_output; /**< Destination tensor */
+};
+
+/** Addition operation */
+class CLSaturatedArithmeticOperationKernel : public CLElementwiseOperationKernel
+{
+public:
+    CLSaturatedArithmeticOperationKernel()
+        : CLElementwiseOperationKernel(), _policy(), _op()
+    {
+    }
+
+    /** Static function to check if given info will lead to a valid configuration of @ref CLSaturatedArithmeticOperationKernel
+     *
+     * @param[in] op     Arithmetic operation to be executed.
+     * @param[in] input1 First tensor input. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input1 First tensor input. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input2 Second tensor input. Data types supported: Same as @p input1.
+     * @param[in] output Output tensor. Data types supported: Same as @p input1.
+     * @param[in] policy Policy to use to handle overflow.
+     */
+    void configure(ArithmeticOperation op, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output, const ConvertPolicy &policy);
+
+    /** Static function to check if given info will lead to a valid configuration of @ref CLSaturatedArithmeticOperationKernel
+     *
+     * @param[in] op     Arithmetic operation to be executed.
+     * @param[in] input1 First tensor input info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input1 First tensor input info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input2 Second tensor input info. Data types supported: Same as @p input1.
+     * @param[in] output Output tensor info. Data types supported: Same as @p input1.
+     * @param[in] policy Policy to use to handle overflow.
+     *
+     * @return a Status
+     */
+    static Status validate(ArithmeticOperation op, const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output, const ConvertPolicy &policy);
+
+protected:
+    // Inherited methods overridden:
+    std::string name() override;
+    std::pair<Status, Window> validate_and_configure_window(ITensorInfo &input1, ITensorInfo &input2, ITensorInfo &output) override;
+    Status validate_arguments(const ITensorInfo &input1, const ITensorInfo &input2, const ITensorInfo &output) override;
+    CLBuildOptions generate_build_options(const ITensorInfo &input1, const ITensorInfo &input2, const ITensorInfo &output) override;
+    std::string generate_id_for_tuning(const std::string &kernel_name, const ITensorInfo &input1, const ITensorInfo &output) override;
+
+private:
+    ConvertPolicy       _policy;
+    ArithmeticOperation _op;
+};
+
+class CLArithmeticOperationKernel : public CLElementwiseOperationKernel
+{
+public:
+    CLArithmeticOperationKernel()
+        : CLElementwiseOperationKernel(), _op()
+    {
+    }
+
+    /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticOperationKernel
+     *
+     * @param[in] op     Arithmetic operation to be executed.
+     * @param[in] input1 First tensor input. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input1 First tensor input. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input2 Second tensor input. Data types supported: Same as @p input1.
+     * @param[in] output Output tensor. Data types supported: Same as @p input1.
+     */
+    void configure(ArithmeticOperation op, const ICLTensor *input1, const ICLTensor *input2, ICLTensor *output);
+
+    /** Static function to check if given info will lead to a valid configuration of @ref CLArithmeticOperationKernel
+     *
+     * @param[in] op     Arithmetic operation to be executed.
+     * @param[in] input1 First tensor input info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input1 First tensor input info. Data types supported: U8/S8/QASYMM8/U16/S16/F16/U32/S32/F32.
+     * @param[in] input2 Second tensor input info. Data types supported: Same as @p input1.
+     * @param[in] output Output tensor info. Data types supported: Same as @p input1.
+     *
+     * @return a Status
+     */
+    static Status validate(ArithmeticOperation op, const ITensorInfo *input1, const ITensorInfo *input2, const ITensorInfo *output);
+
+protected:
+    // Inherited methods overridden:
+    std::string name() override;
+    std::pair<Status, Window> validate_and_configure_window(ITensorInfo &input1, ITensorInfo &input2, ITensorInfo &output) override;
+    Status validate_arguments(const ITensorInfo &input1, const ITensorInfo &input2, const ITensorInfo &output) override;
+    CLBuildOptions generate_build_options(const ITensorInfo &input1, const ITensorInfo &input2, const ITensorInfo &output) override;
+    std::string generate_id_for_tuning(const std::string &kernel_name, const ITensorInfo &input1, const ITensorInfo &output) override;
+
+private:
+    ArithmeticOperation _op;
+};
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_CLELEMENTWISEOPERATIONKERNEL_H__ */