Make CL Elementwise Unary kernels and functions state-less

Resolves COMPMID-4004

Change-Id: I1dfe8bc52c1ff394ea208ba98b51033c738746a4
Signed-off-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/4922
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

13 files changed, 532 insertions, 437 deletions

diff --git a/Android.bp b/Android.bp
index a1898b5432..663adaedfc 100644
--- a/Android.bp
+++ b/Android.bp
@@ -109,7 +109,6 @@ cc_library_static {
         "src/core/CL/kernels/CLDerivativeKernel.cpp",
         "src/core/CL/kernels/CLDilateKernel.cpp",
         "src/core/CL/kernels/CLDirectConvolutionLayerKernel.cpp",
-        "src/core/CL/kernels/CLElementWiseUnaryLayerKernel.cpp",
         "src/core/CL/kernels/CLErodeKernel.cpp",
         "src/core/CL/kernels/CLFFTDigitReverseKernel.cpp",
         "src/core/CL/kernels/CLFFTRadixStageKernel.cpp",
@@ -441,6 +440,7 @@ cc_library_static {
         "src/core/gpu/cl/kernels/ClBatchConcatenateKernel.cpp",
         "src/core/gpu/cl/kernels/ClDepthConcatenateKernel.cpp",
         "src/core/gpu/cl/kernels/ClElementwiseKernel.cpp",
+        "src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.cpp",
         "src/core/gpu/cl/kernels/ClFloorKernel.cpp",
         "src/core/gpu/cl/kernels/ClHeightConcatenateKernel.cpp",
         "src/core/gpu/cl/kernels/ClWidthConcatenate2TensorsKernel.cpp",
@@ -798,7 +798,9 @@ cc_library_static {
         "src/runtime/gpu/cl/operators/ClAdd.cpp",
         "src/runtime/gpu/cl/operators/ClConcatenate.cpp",
         "src/runtime/gpu/cl/operators/ClElementwiseOperations.cpp",
+        "src/runtime/gpu/cl/operators/ClElementwiseUnary.cpp",
         "src/runtime/gpu/cl/operators/ClFloor.cpp",
+        "src/runtime/gpu/cl/operators/ClLogicalNot.cpp",
         "src/runtime/gpu/cl/operators/ClSub.cpp",
         "utils/CommonGraphOptions.cpp",
         "utils/GraphUtils.cpp",
diff --git a/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h b/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h
index 72b5b7dee8..fd6942cad5 100644
--- a/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h
+++ b/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -24,8 +24,11 @@
 #ifndef ARM_COMPUTE_CLELEMENTWISEUNARYLAYER_H
 #define ARM_COMPUTE_CLELEMENTWISEUNARYLAYER_H
 
-#include "arm_compute/runtime/CL/ICLOperator.h"
-#include "arm_compute/runtime/CL/ICLSimpleFunction.h"
+#include "arm_compute/runtime/IFunction.h"
+
+#include "arm_compute/core/Types.h"
+
+#include <memory>
 
 namespace arm_compute
 {
@@ -354,197 +357,5 @@ private:
     struct Impl;
     std::unique_ptr<Impl> _impl;
 };
-
-namespace experimental
-{
-/** Basic function to perform inverse square root on an input tensor. */
-class CLRsqrt : public ICLOperator
-{
-public:
-    /** Initialize the function
-     *
-     * @param[in]  input  Input tensor info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output);
-    /** Initialize the function
-     *
-     * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           Input tensor info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLRsqrtLayer
-     *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-};
-
-/** Basic function to perform exponential on an input tensor. */
-class CLExp : public ICLOperator
-{
-public:
-    /** Initialize the function
-     *
-     * @param[in]  input  Input tensor info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output);
-    /** Initialize the function
-     *
-     * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           Input tensor info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLExpLayer
-     *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-};
-
-/** Basic function to negate an input tensor. */
-class CLNeg : public ICLOperator
-{
-public:
-    /** Initialize the function
-     *
-     * @param[in]  input  Input tensor info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output);
-    /** Initialize the function
-     *
-     * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           Input tensor info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLNegLayer
-     *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-};
-
-/** Basic function to calculate sine of an input tensor. */
-class CLSin : public ICLOperator
-{
-public:
-    /** Initialize the function
-     *
-     * @param[in]  input  Input tensor info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output);
-    /** Initialize the function
-     *
-     * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           Input tensor info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLSinLayer
-     *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-};
-
-/** Basic function to perform elementwise log on an input tensor. */
-class CLLog : public ICLOperator
-{
-public:
-    /** Initialize the function
-     *
-     * @param[in]  input  Input tensor info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output);
-    /** Initialize the function
-     *
-     * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           Input tensor info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLLogLayer
-     *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-};
-
-/** Basic function to get the absolute value of an input tensor. */
-class CLAbs : public ICLOperator
-{
-public:
-    /** Initialize the function
-     *
-     * @param[in]  input  Input tensor info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output);
-    /** Initialize the function
-     *
-     * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           Input tensor info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLAbsLayer
-     *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-};
-
-/** Basic function to get the round (to the nearest even) value of an input tensor. */
-class CLRound : public ICLOperator
-{
-public:
-    /** Initialize the function
-     *
-     * @param[in]  input  Input tensor info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output);
-    /** Initialize the function
-     *
-     * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           Input tensor info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLRoundLayer
-     *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-};
-} // namespace experimental
 } // namespace arm_compute
 #endif /* ARM_COMPUTE_CLELEMENTWISEUNARYLAYER_H */
diff --git a/arm_compute/runtime/CL/functions/CLLogicalNot.h b/arm_compute/runtime/CL/functions/CLLogicalNot.h
index f1f2ac1162..772f16b942 100644
--- a/arm_compute/runtime/CL/functions/CLLogicalNot.h
+++ b/arm_compute/runtime/CL/functions/CLLogicalNot.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020 Arm Limited.
+ * Copyright (c) 2020-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -24,43 +24,18 @@
 #ifndef ARM_COMPUTE_CLLOGICALNOT_H
 #define ARM_COMPUTE_CLLOGICALNOT_H
 
-#include "arm_compute/core/Error.h"
-#include "arm_compute/runtime/CL/ICLOperator.h"
 #include "arm_compute/runtime/IFunction.h"
 
+#include "arm_compute/core/Types.h"
+
+#include <memory>
+
 namespace arm_compute
 {
 class CLCompileContext;
 class ICLTensor;
 class ITensorInfo;
 
-namespace experimental
-{
-class CLLogicalNot : public ICLOperator
-{
-public:
-    /** Default Constructor */
-    CLLogicalNot() = default;
-    /** Initialise the kernel's inputs, output and conversion policy.
-     *
-     * @param[in]      compile_context The compile context to be used.
-     * @param[in, out] input           Tensor input. Data types supported: U8.
-     * @param[out]     output          Output tensor. Data types supported: same as @p input.
-     */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLElementWiseUnaryLayerKernel
-     *
-     * @param[in] input  Tensor input info. Data types supported: U8.
-     * @param[in] output Output tensor info. Data types supported: same as @p input.
-     *
-     * @return a status
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
-    // Inherited methods overridden:
-    void run(ITensorPack &tensors) override;
-};
-} // namespace experimental
-
 /** Basic function to do logical NOT operation
  *
  * @note The tensor data type for the inputs must be U8.
diff --git a/docs/00_introduction.dox b/docs/00_introduction.dox
index d7ef3dcb7d..ccc44d48e0 100644
--- a/docs/00_introduction.dox
+++ b/docs/00_introduction.dox
@@ -856,7 +856,7 @@ v19.02 Public major release
     - @ref CLReverse / @ref CLReverseKernel
     - @ref CLRsqrtLayer
     - @ref CLExpLayer
-    - @ref CLElementWiseUnaryLayerKernel
+    - CLElementWiseUnaryLayerKernel
     - @ref CLGEMMReshapeLHSMatrixKernel
     - @ref CLGEMMReshapeRHSMatrixKernel
     - @ref CLGEMMMatrixMultiplyReshapedKernel
diff --git a/src/core/CL/CLKernels.h b/src/core/CL/CLKernels.h
index 916cf456c1..d31c876881 100644
--- a/src/core/CL/CLKernels.h
+++ b/src/core/CL/CLKernels.h
@@ -56,7 +56,6 @@
 #include "src/core/CL/kernels/CLDerivativeKernel.h"
 #include "src/core/CL/kernels/CLDilateKernel.h"
 #include "src/core/CL/kernels/CLDirectConvolutionLayerKernel.h"
-#include "src/core/CL/kernels/CLElementWiseUnaryLayerKernel.h"
 #include "src/core/CL/kernels/CLErodeKernel.h"
 #include "src/core/CL/kernels/CLFFTDigitReverseKernel.h"
 #include "src/core/CL/kernels/CLFFTRadixStageKernel.h"
diff --git a/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.cpp b/src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.cpp
index e7c95ba09e..8e9698cd50 100644
--- a/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.cpp
+++ b/src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -21,7 +21,7 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#include "src/core/CL/kernels/CLElementWiseUnaryLayerKernel.h"
+#include "src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.h"
 
 #include "arm_compute/core/CL/CLHelpers.h"
 #include "arm_compute/core/CL/ICLTensor.h"
@@ -32,55 +32,54 @@
 
 namespace arm_compute
 {
+namespace opencl
+{
+namespace kernels
+{
 namespace
 {
-Status validate_arguments(const ITensorInfo &input, const ITensorInfo &output, const ElementWiseUnary op)
+Status validate_arguments(const ITensorInfo &src, const ITensorInfo &dst, const ElementWiseUnary op)
 {
-    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&input);
+    ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&src);
     if(op == ElementWiseUnary::LOGICAL_NOT)
     {
-        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&input, 1, DataType::U8);
+        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&src, 1, DataType::U8);
     }
     else
     {
-        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&input, 1, DataType::F16, DataType::F32);
+        ARM_COMPUTE_RETURN_ERROR_ON_DATA_TYPE_CHANNEL_NOT_IN(&src, 1, DataType::F16, DataType::F32);
     }
 
-    // Validate in case of configured output
-    if(output.total_size() > 0)
+    // Validate in case of configured dst
+    if(dst.total_size() > 0)
     {
-        ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&output);
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&input, &output);
-        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(&input, &output);
+        ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(&dst);
+        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(&src, &dst);
+        ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_SHAPES(&src, &dst);
     }
 
     return Status{};
 }
 } // namespace
 
-void CLElementWiseUnaryLayerKernel::configure(const ITensorInfo *input, ITensorInfo *output, const ElementWiseUnary &op)
-{
-    configure(CLKernelLibrary::get().get_compile_context(), input, output, op);
-}
-
-void CLElementWiseUnaryLayerKernel::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output, const ElementWiseUnary &op)
+void ClElementWiseUnaryKernel::configure(const CLCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst, const ElementWiseUnary &op)
 {
-    ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
+    ARM_COMPUTE_ERROR_ON_NULLPTR(src, dst);
 
-    auto padding_info = get_padding_info({ input, output });
+    auto padding_info = get_padding_info({ src, dst });
 
-    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(*input, *output, op));
+    ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(*src, *dst, op));
 
     const std::string kernel_name    = "elementwise_unary";
-    const int         vec_size_x     = 16 / output->element_size();
-    const int         output_width_x = output->tensor_shape().x();
-    const bool        multi_access_x = (output_width_x / vec_size_x > 0);
+    const int         vec_size_x     = 16 / dst->element_size();
+    const int         dst_width_x    = dst->tensor_shape().x();
+    const bool        multi_access_x = (dst_width_x / vec_size_x > 0);
 
     // Set kernel build options
     CLBuildOptions build_opts;
-    build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(input->data_type()));
+    build_opts.add_option("-DDATA_TYPE=" + get_cl_type_from_data_type(src->data_type()));
     build_opts.add_option_if(multi_access_x, "-DVEC_SIZE=" + support::cpp11::to_string(vec_size_x));
-    build_opts.add_option_if(multi_access_x, "-DLAST_ACCESSED_X=" + support::cpp11::to_string(std::max<int>(output_width_x - vec_size_x, 0)));
+    build_opts.add_option_if(multi_access_x, "-DLAST_ACCESSED_X=" + support::cpp11::to_string(std::max<int>(dst_width_x - vec_size_x, 0)));
     switch(op)
     {
         case ElementWiseUnary::RSQRT:
@@ -115,7 +114,7 @@ void CLElementWiseUnaryLayerKernel::configure(const CLCompileContext &compile_co
     _kernel = create_kernel(compile_context, kernel_name, build_opts.options());
 
     // Configure kernel window
-    Window win = calculate_max_window(*output);
+    Window win = calculate_max_window(*dst);
     if(multi_access_x)
     {
         win.set(Window::DimX,
@@ -126,16 +125,16 @@ void CLElementWiseUnaryLayerKernel::configure(const CLCompileContext &compile_co
     ARM_COMPUTE_ERROR_ON(has_padding_changed(padding_info));
 }
 
-Status CLElementWiseUnaryLayerKernel::validate(const ITensorInfo *input, const ITensorInfo *output, const ElementWiseUnary &op)
+Status ClElementWiseUnaryKernel::validate(const ITensorInfo *src, const ITensorInfo *dst, const ElementWiseUnary &op)
 {
     ARM_COMPUTE_UNUSED(op);
-    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
-    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(*input, *output, op));
+    ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(src, dst);
+    ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(*src, *dst, op));
 
     return Status{};
 }
 
-void CLElementWiseUnaryLayerKernel::run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue)
+void ClElementWiseUnaryKernel::run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue)
 {
     ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
     ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
@@ -155,4 +154,6 @@ void CLElementWiseUnaryLayerKernel::run_op(ITensorPack &tensors, const Window &w
     }
     while(collapsed.slide_window_slice_3D(slice));
 }
+} // namespace kernels
+} // namespace opencl
 } // namespace arm_compute
diff --git a/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.h b/src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.h
index 95b5872796..7e5edef3ee 100644
--- a/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.h
+++ b/src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -21,46 +21,47 @@
  * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  * SOFTWARE.
  */
-#ifndef ARM_COMPUTE_CLELEMENTWISEUNARYLAYERKERNEL_H
-#define ARM_COMPUTE_CLELEMENTWISEUNARYLAYERKERNEL_H
+#ifndef ARM_COMPUTE_CL_ELEMENTWISE_UNARY_KERNEL_H
+#define ARM_COMPUTE_CL_ELEMENTWISE_UNARY_KERNEL_H
 
-#include "arm_compute/core/Types.h"
-#include "src/core/CL/ICLKernel.h"
-#include "src/core/CL/ICLSimpleKernel.h"
+#include "src/core/common/Macros.h"
+#include "src/core/gpu/cl/ClCompileContext.h"
+#include "src/core/gpu/cl/IClKernel.h"
 
 namespace arm_compute
 {
+namespace opencl
+{
+namespace kernels
+{
 /** Interface for the elementwise unary operator */
-class CLElementWiseUnaryLayerKernel : public ICLKernel
+class ClElementWiseUnaryKernel : public IClKernel
 {
 public:
-    /** Initialise the kernel's inputs, output.
-     *
-     * @param[in]  input  First tensor input info. Data types supported: F16/F32.
-     * @param[out] output Output tensor info. Data types supported: Same as @p input.
-     * @param[in]  op     Element wise unary operation to perform.
-     */
-    void configure(const ITensorInfo *input, ITensorInfo *output, const ElementWiseUnary &op);
-    /** Initialise the kernel's inputs, output.
+    ClElementWiseUnaryKernel() = default;
+    ARM_COMPUTE_DISALLOW_COPY_ALLOW_MOVE(ClElementWiseUnaryKernel);
+    /** Initialise the kernel's srcs, dst.
      *
      * @param[in]  compile_context The compile context to be used.
-     * @param[in]  input           First tensor input info. Data types supported: F16/F32.
-     * @param[out] output          Output tensor info. Data types supported: Same as @p input.
+     * @param[in]  src             First source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
      * @param[in]  op              Element wise unary operation to perform.
      */
-    void configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output, const ElementWiseUnary &op);
-    /** Static function to check if given info will lead to a valid configuration of @ref CLElementWiseUnaryLayerKernel
+    void configure(const CLCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst, const ElementWiseUnary &op);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClElementWiseUnaryKernel
      *
-     * @param[in] input  First tensor input info. Data types supported: F16/F32.
-     * @param[in] output Output tensor info. Data types supported: Same as @p input.
-     * @param[in] op     Element wise unary operation to perform.
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     * @param[in] op  Element wise unary operation to perform.
      *
      * @return a status
      */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output, const ElementWiseUnary &op);
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst, const ElementWiseUnary &op);
 
     // Inherited methods overridden:
     void run_op(ITensorPack &tensors, const Window &window, cl::CommandQueue &queue) override;
 };
+} // namespace kernels
+} // namespace opencl
 } // namespace arm_compute
-#endif /* ARM_COMPUTE_CLELEMENTWISEUNARYLAYERKERNEL_H */
+#endif /* ARM_COMPUTE_CL_ELEMENTWISE_UNARY_KERNEL_H */
diff --git a/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp b/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp
index 0ded640f51..dbf2121bfd 100644
--- a/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp
+++ b/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -23,104 +23,18 @@
  */
 #include "arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h"
 
-#include "src/core/CL/kernels/CLElementWiseUnaryLayerKernel.h"
-
-#include <utility>
+#include "arm_compute/core/CL/CLKernelLibrary.h"
+#include "arm_compute/core/CL/ICLTensor.h"
+#include "src/core/CL/ICLKernel.h"
+#include "src/runtime/gpu/cl/operators/ClElementwiseUnary.h"
 
 namespace arm_compute
 {
-namespace experimental
-{
-void CLRsqrt::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::RSQRT);
-    _kernel = std::move(k);
-}
-
-Status CLRsqrt::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return arm_compute::CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::RSQRT);
-}
-
-void CLExp::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::EXP);
-    _kernel = std::move(k);
-}
-
-Status CLExp::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return arm_compute::CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::EXP);
-}
-
-void CLNeg::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::NEG);
-    _kernel = std::move(k);
-}
-
-Status CLNeg::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return arm_compute::CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::NEG);
-}
-
-void CLSin::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::SIN);
-    _kernel = std::move(k);
-}
-
-Status CLSin::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return arm_compute::CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::SIN);
-}
-
-void CLAbs::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::ABS);
-    _kernel = std::move(k);
-}
-
-Status CLAbs::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return arm_compute::CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::ABS);
-}
-
-void CLLog::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::LOG);
-    _kernel = std::move(k);
-}
-
-Status CLLog::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return arm_compute::CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::LOG);
-}
-
-void CLRound::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::ROUND);
-    _kernel = std::move(k);
-}
-
-Status CLRound::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return arm_compute::CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::ROUND);
-}
-} // namespace experimental
-
 struct CLRsqrtLayer::Impl
 {
-    const ICLTensor                       *src{ nullptr };
-    ICLTensor                             *dst{ nullptr };
-    std::unique_ptr<experimental::CLRsqrt> op{ nullptr };
+    const ICLTensor                 *src{ nullptr };
+    ICLTensor                       *dst{ nullptr };
+    std::unique_ptr<opencl::ClRsqrt> op{ nullptr };
 };
 
 CLRsqrtLayer::CLRsqrtLayer()
@@ -141,13 +55,13 @@ void CLRsqrtLayer::configure(const CLCompileContext &compile_context, const ICLT
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLRsqrt>();
+    _impl->op  = std::make_unique<opencl::ClRsqrt>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 
 Status CLRsqrtLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLRsqrt::validate(input, output);
+    return opencl::ClRsqrt::validate(input, output);
 }
 
 void CLRsqrtLayer::run()
@@ -160,9 +74,9 @@ void CLRsqrtLayer::run()
 
 struct CLExpLayer::Impl
 {
-    const ICLTensor                     *src{ nullptr };
-    ICLTensor                           *dst{ nullptr };
-    std::unique_ptr<experimental::CLExp> op{ nullptr };
+    const ICLTensor               *src{ nullptr };
+    ICLTensor                     *dst{ nullptr };
+    std::unique_ptr<opencl::ClExp> op{ nullptr };
 };
 
 CLExpLayer::CLExpLayer()
@@ -183,13 +97,13 @@ void CLExpLayer::configure(const CLCompileContext &compile_context, const ICLTen
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLExp>();
+    _impl->op  = std::make_unique<opencl::ClExp>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 
 Status CLExpLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLExp::validate(input, output);
+    return opencl::ClExp::validate(input, output);
 }
 
 void CLExpLayer::run()
@@ -202,9 +116,9 @@ void CLExpLayer::run()
 
 struct CLNegLayer::Impl
 {
-    const ICLTensor                     *src{ nullptr };
-    ICLTensor                           *dst{ nullptr };
-    std::unique_ptr<experimental::CLNeg> op{ nullptr };
+    const ICLTensor               *src{ nullptr };
+    ICLTensor                     *dst{ nullptr };
+    std::unique_ptr<opencl::ClNeg> op{ nullptr };
 };
 
 CLNegLayer::CLNegLayer()
@@ -225,12 +139,12 @@ void CLNegLayer::configure(const CLCompileContext &compile_context, const ICLTen
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLNeg>();
+    _impl->op  = std::make_unique<opencl::ClNeg>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 Status CLNegLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLNeg::validate(input, output);
+    return opencl::ClNeg::validate(input, output);
 }
 
 void CLNegLayer::run()
@@ -243,9 +157,9 @@ void CLNegLayer::run()
 
 struct CLSinLayer::Impl
 {
-    const ICLTensor                     *src{ nullptr };
-    ICLTensor                           *dst{ nullptr };
-    std::unique_ptr<experimental::CLSin> op{ nullptr };
+    const ICLTensor               *src{ nullptr };
+    ICLTensor                     *dst{ nullptr };
+    std::unique_ptr<opencl::ClSin> op{ nullptr };
 };
 
 CLSinLayer::CLSinLayer()
@@ -266,12 +180,12 @@ void CLSinLayer::configure(const CLCompileContext &compile_context, const ICLTen
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLSin>();
+    _impl->op  = std::make_unique<opencl::ClSin>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 Status CLSinLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLSin::validate(input, output);
+    return opencl::ClSin::validate(input, output);
 }
 
 void CLSinLayer::run()
@@ -284,9 +198,9 @@ void CLSinLayer::run()
 
 struct CLAbsLayer::Impl
 {
-    const ICLTensor                     *src{ nullptr };
-    ICLTensor                           *dst{ nullptr };
-    std::unique_ptr<experimental::CLAbs> op{ nullptr };
+    const ICLTensor               *src{ nullptr };
+    ICLTensor                     *dst{ nullptr };
+    std::unique_ptr<opencl::ClAbs> op{ nullptr };
 };
 
 CLAbsLayer::CLAbsLayer()
@@ -307,12 +221,12 @@ void CLAbsLayer::configure(const CLCompileContext &compile_context, const ICLTen
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLAbs>();
+    _impl->op  = std::make_unique<opencl::ClAbs>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 Status CLAbsLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLAbs::validate(input, output);
+    return opencl::ClAbs::validate(input, output);
 }
 
 void CLAbsLayer::run()
@@ -325,9 +239,9 @@ void CLAbsLayer::run()
 
 struct CLLogLayer::Impl
 {
-    const ICLTensor                     *src{ nullptr };
-    ICLTensor                           *dst{ nullptr };
-    std::unique_ptr<experimental::CLLog> op{ nullptr };
+    const ICLTensor               *src{ nullptr };
+    ICLTensor                     *dst{ nullptr };
+    std::unique_ptr<opencl::ClLog> op{ nullptr };
 };
 
 CLLogLayer::CLLogLayer()
@@ -348,12 +262,12 @@ void CLLogLayer::configure(const CLCompileContext &compile_context, const ICLTen
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLLog>();
+    _impl->op  = std::make_unique<opencl::ClLog>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 Status CLLogLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLLog::validate(input, output);
+    return opencl::ClLog::validate(input, output);
 }
 
 void CLLogLayer::run()
@@ -366,9 +280,9 @@ void CLLogLayer::run()
 
 struct CLRoundLayer::Impl
 {
-    const ICLTensor                       *src{ nullptr };
-    ICLTensor                             *dst{ nullptr };
-    std::unique_ptr<experimental::CLRound> op{ nullptr };
+    const ICLTensor                 *src{ nullptr };
+    ICLTensor                       *dst{ nullptr };
+    std::unique_ptr<opencl::ClRound> op{ nullptr };
 };
 
 CLRoundLayer::CLRoundLayer()
@@ -389,12 +303,12 @@ void CLRoundLayer::configure(const CLCompileContext &compile_context, const ICLT
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLRound>();
+    _impl->op  = std::make_unique<opencl::ClRound>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 Status CLRoundLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLRound::validate(input, output);
+    return opencl::ClRound::validate(input, output);
 }
 
 void CLRoundLayer::run()
diff --git a/src/runtime/CL/functions/CLLogicalNot.cpp b/src/runtime/CL/functions/CLLogicalNot.cpp
index d3774da597..388d2bce86 100644
--- a/src/runtime/CL/functions/CLLogicalNot.cpp
+++ b/src/runtime/CL/functions/CLLogicalNot.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020 Arm Limited.
+ * Copyright (c) 2020-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -22,38 +22,19 @@
  * SOFTWARE.
  */
 #include "arm_compute/runtime/CL/functions/CLLogicalNot.h"
-#include "arm_compute/core/CL/ICLTensor.h"
-#include "src/core/CL/kernels/CLElementWiseUnaryLayerKernel.h"
 
-#include <utility>
+#include "arm_compute/core/CL/CLKernelLibrary.h"
+#include "arm_compute/core/CL/ICLTensor.h"
+#include "src/core/CL/ICLKernel.h"
+#include "src/runtime/gpu/cl/operators/ClLogicalNot.h"
 
 namespace arm_compute
 {
-namespace experimental
-{
-void CLLogicalNot::configure(const CLCompileContext &compile_context, const ITensorInfo *input, ITensorInfo *output)
-{
-    auto k = std::make_unique<CLElementWiseUnaryLayerKernel>();
-    k->configure(compile_context, input, output, ElementWiseUnary::LOGICAL_NOT);
-    _kernel = std::move(k);
-}
-
-Status CLLogicalNot::validate(const ITensorInfo *input, const ITensorInfo *output)
-{
-    return CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::LOGICAL_NOT);
-}
-
-void CLLogicalNot::run(ITensorPack &tensors)
-{
-    ICLOperator::run(tensors);
-}
-} // namespace experimental
-
 struct CLLogicalNot::Impl
 {
-    const ICLTensor                            *src{ nullptr };
-    ICLTensor                                  *dst{ nullptr };
-    std::unique_ptr<experimental::CLLogicalNot> op{ nullptr };
+    const ICLTensor                      *src{ nullptr };
+    ICLTensor                            *dst{ nullptr };
+    std::unique_ptr<opencl::ClLogicalNot> op{ nullptr };
 };
 
 CLLogicalNot::CLLogicalNot()
@@ -73,13 +54,13 @@ void CLLogicalNot::configure(const CLCompileContext &compile_context, const ICLT
 {
     _impl->src = input;
     _impl->dst = output;
-    _impl->op  = std::make_unique<experimental::CLLogicalNot>();
+    _impl->op  = std::make_unique<opencl::ClLogicalNot>();
     _impl->op->configure(compile_context, input->info(), output->info());
 }
 
 Status CLLogicalNot::validate(const ITensorInfo *input, const ITensorInfo *output)
 {
-    return experimental::CLLogicalNot::validate(input, output);
+    return opencl::ClLogicalNot::validate(input, output);
 }
 
 void CLLogicalNot::run()
diff --git a/src/runtime/gpu/cl/operators/ClElementwiseUnary.cpp b/src/runtime/gpu/cl/operators/ClElementwiseUnary.cpp
new file mode 100644
index 0000000000..7b830a077f
--- /dev/null
+++ b/src/runtime/gpu/cl/operators/ClElementwiseUnary.cpp
@@ -0,0 +1,116 @@
+/*
+ * 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.
+ */
+#include "src/runtime/gpu/cl/operators/ClElementwiseUnary.h"
+
+#include "src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.h"
+
+namespace arm_compute
+{
+namespace opencl
+{
+void ClRsqrt::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::RSQRT);
+    _kernel = std::move(k);
+}
+
+Status ClRsqrt::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::RSQRT);
+}
+
+void ClExp::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::EXP);
+    _kernel = std::move(k);
+}
+
+Status ClExp::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::EXP);
+}
+
+void ClNeg::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::NEG);
+    _kernel = std::move(k);
+}
+
+Status ClNeg::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::NEG);
+}
+
+void ClSin::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::SIN);
+    _kernel = std::move(k);
+}
+
+Status ClSin::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::SIN);
+}
+
+void ClAbs::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::ABS);
+    _kernel = std::move(k);
+}
+
+Status ClAbs::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::ABS);
+}
+
+void ClLog::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::LOG);
+    _kernel = std::move(k);
+}
+
+Status ClLog::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::LOG);
+}
+
+void ClRound::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::ROUND);
+    _kernel = std::move(k);
+}
+
+Status ClRound::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::ROUND);
+}
+} // namespace opencl
+} // namespace arm_compute
diff --git a/src/runtime/gpu/cl/operators/ClElementwiseUnary.h b/src/runtime/gpu/cl/operators/ClElementwiseUnary.h
new file mode 100644
index 0000000000..b40e3e9a3b
--- /dev/null
+++ b/src/runtime/gpu/cl/operators/ClElementwiseUnary.h
@@ -0,0 +1,192 @@
+/*
+ * 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 ARM_COMPUTE_CL_ELEMENTWISE_UNARY_H
+#define ARM_COMPUTE_CL_ELEMENTWISE_UNARY_H
+
+#include "src/core/gpu/cl/ClCompileContext.h"
+#include "src/runtime/gpu/cl/IClOperator.h"
+
+namespace arm_compute
+{
+namespace opencl
+{
+/** Basic function to perform inverse square root on an src tensor. */
+class ClRsqrt : public IClOperator
+{
+public:
+    /** Constructor */
+    ClRsqrt() = default;
+    /** Initialize the function
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClRsqrt
+     *
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+
+/** Basic function to perform exponential on an src tensor. */
+class ClExp : public IClOperator
+{
+public:
+    /** Constructor */
+    ClExp() = default;
+    /** Initialize the function
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClExp
+     *
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+
+/** Basic function to negate an src tensor. */
+class ClNeg : public IClOperator
+{
+public:
+    /** Constructor */
+    ClNeg() = default;
+    /** Initialize the function
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClNeg
+     *
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+
+/** Basic function to calculate sine of an src tensor. */
+class ClSin : public IClOperator
+{
+public:
+    /** Constructor */
+    ClSin() = default;
+    /** Initialize the function
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClSin
+     *
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+
+/** Basic function to perform elementwise log on an src tensor. */
+class ClLog : public IClOperator
+{
+public:
+    /** Constructor */
+    ClLog() = default;
+    /** Initialize the function
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClLog
+     *
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+
+/** Basic function to get the absolute value of an src tensor. */
+class ClAbs : public IClOperator
+{
+public:
+    /** Initialize the function
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClAbs
+     *
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+
+/** Basic function to get the round (to the nearest even) value of an src tensor. */
+class ClRound : public IClOperator
+{
+public:
+    /** Initialize the function
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: F16/F32.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration of @ref ClRound
+     *
+     * @param[in] src First source tensor info. Data types supported: F16/F32.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+} // namespace opencl
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_CL_ELEMENTWISE_UNARY_H */
diff --git a/src/runtime/gpu/cl/operators/ClLogicalNot.cpp b/src/runtime/gpu/cl/operators/ClLogicalNot.cpp
new file mode 100644
index 0000000000..400efe450d
--- /dev/null
+++ b/src/runtime/gpu/cl/operators/ClLogicalNot.cpp
@@ -0,0 +1,45 @@
+/*
+ * Copyright (c) 2017-2021 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.
+ */
+#include "src/runtime/gpu/cl/operators/ClLogicalNot.h"
+
+#include "src/core/gpu/cl/ClCompileContext.h"
+#include "src/core/gpu/cl/kernels/ClElementwiseUnaryKernel.h"
+
+namespace arm_compute
+{
+namespace opencl
+{
+void ClLogicalNot::configure(const ClCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst)
+{
+    auto k = std::make_unique<kernels::ClElementWiseUnaryKernel>();
+    k->configure(compile_context, src, dst, ElementWiseUnary::LOGICAL_NOT);
+    _kernel = std::move(k);
+}
+
+Status ClLogicalNot::validate(const ITensorInfo *src, const ITensorInfo *dst)
+{
+    return kernels::ClElementWiseUnaryKernel::validate(src, dst, ElementWiseUnary::LOGICAL_NOT);
+}
+} // namespace opencl
+} // namespace arm_compute
diff --git a/src/runtime/gpu/cl/operators/ClLogicalNot.h b/src/runtime/gpu/cl/operators/ClLogicalNot.h
new file mode 100644
index 0000000000..25ddf564b5
--- /dev/null
+++ b/src/runtime/gpu/cl/operators/ClLogicalNot.h
@@ -0,0 +1,58 @@
+/*
+ * 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 ARM_COMPUTE_CL_LOGICAL_NOT_H
+#define ARM_COMPUTE_CL_LOGICAL_NOT_H
+
+#include "src/core/gpu/cl/ClCompileContext.h"
+#include "src/runtime/gpu/cl/IClOperator.h"
+
+namespace arm_compute
+{
+namespace opencl
+{
+/** Basic function to run @ref kernels::ClElementWiseUnaryKernel for NOT operation */
+class ClLogicalNot : public IClOperator
+{
+public:
+    /** Constructor */
+    ClLogicalNot() = default;
+    /** Configure operator for a given list of arguments
+     *
+     * @param[in]  compile_context The compile context to be used.
+     * @param[in]  src             Source tensor info. Data types supported: U8.
+     * @param[out] dst             Destination tensor info. Data types supported: same as @p src.
+     */
+    void configure(const CLCompileContext &compile_context, const ITensorInfo *src, ITensorInfo *dst);
+    /** Static function to check if given info will lead to a valid configuration
+     *
+     * @param[in] src Soure tensor info. Data types supported: U8.
+     * @param[in] dst Destination tensor info. Data types supported: same as @p src.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *src, const ITensorInfo *dst);
+};
+} // namespace opencl
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_CL_LOGICAL_NOT_H */