diff --git a/arm_compute/core/CPP/ICPPSimpleKernel.h b/arm_compute/core/CPP/ICPPSimpleKernel.h
deleted file mode 100644
index c31d487..0000000
--- a/arm_compute/core/CPP/ICPPSimpleKernel.h
+++ /dev/null
@@ -1,76 +0,0 @@
-/*
- * Copyright (c) 2017-2019 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_ICPPSIMPLEKERNEL_H
-#define ARM_COMPUTE_ICPPSIMPLEKERNEL_H
-
-#include "arm_compute/core/CPP/ICPPKernel.h"
-
-namespace arm_compute
-{
-class ITensor;
-
-/** Interface for simple C++ kernels having 1 tensor input and 1 tensor output */
-class ICPPSimpleKernel : public ICPPKernel
-{
-public:
-    /** Constructor */
-    ICPPSimpleKernel();
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    ICPPSimpleKernel(const ICPPSimpleKernel &) = delete;
-    /** Prevent instances of this class from being copied (As this class contains pointers) */
-    ICPPSimpleKernel &operator=(const ICPPSimpleKernel &) = delete;
-    /** Allow instances of this class to be moved */
-    ICPPSimpleKernel(ICPPSimpleKernel &&) = default;
-    /** Allow instances of this class to be moved */
-    ICPPSimpleKernel &operator=(ICPPSimpleKernel &&) = default;
-    /** Default destructor */
-    ~ICPPSimpleKernel() = default;
-
-protected:
-    /** Configure the kernel
-     *
-     * @param[in]  input                             Source tensor.
-     * @param[out] output                            Destination tensor.
-     * @param[in]  num_elems_processed_per_iteration Number of processed elements per iteration.
-     * @param[in]  border_undefined                  (Optional) True if the border mode is undefined. False if it's replicate or constant.
-     * @param[in]  border_size                       (Optional) Size of the border.
-     */
-    void configure(const ITensor *input, ITensor *output, unsigned int num_elems_processed_per_iteration, bool border_undefined = false, const BorderSize &border_size = BorderSize());
-    /** Static function to check if given info will lead to a valid configuration of @ref ICPPSimpleKernel.
-     *
-     * @param[in] input                             Source tensor info.
-     * @param[in] output                            Destination tensor info.
-     * @param[in] num_elems_processed_per_iteration Number of processed elements per iteration.
-     * @param[in] border_undefined                  (Optional) True if the border mode is undefined. False if it's replicate or constant.
-     * @param[in] border_size                       (Optional) Size of the border.
-     */
-    static Status validate(const ITensorInfo *input, const ITensorInfo *output, unsigned int num_elems_processed_per_iteration,
-                           bool border_undefined = false, const BorderSize &border_size = BorderSize());
-
-protected:
-    const ITensor *_input;
-    ITensor       *_output;
-};
-} // namespace arm_compute
-#endif /*ARM_COMPUTE_ICPPSIMPLEKERNEL_H */
diff --git a/arm_compute/core/Types.h b/arm_compute/core/Types.h
index 48c87cd..f6658e7 100644
--- a/arm_compute/core/Types.h
+++ b/arm_compute/core/Types.h
@@ -1750,10 +1750,10 @@
 
 /** GEMM reshape information class. This class stores the necessary information about matrix A and matrix B reshape.
  *
- * The matrix A can only be reshaped through @ref opencl::kernels::ClGemmReshapeLhsMatrixKernel or  @ref NEGEMMInterleave4x4Kernel
+ * The matrix A can only be reshaped through @ref opencl::kernels::ClGemmReshapeLhsMatrixKernel or  @ref cpu::kernels::CpuGemmInterleave4x4Kernel
  * Note: Optionally just for @ref opencl::kernels::ClGemmReshapeLhsMatrixKernel is it possible to set mult_interleave4x4_height, the multiplication factor for the height of the 4x4 interleaved block
  *
- * The matrix B can only be reshaped through @ref opencl::kernels::ClGemmReshapeRhsMatrixKernel or  @ref NEGEMMTranspose1xWKernel
+ * The matrix B can only be reshaped through @ref opencl::kernels::ClGemmReshapeRhsMatrixKernel or  @ref cpu::kernels::CpuGemmTranspose1xWKernel
  * Note: Optionally just for @ref opencl::kernels::ClGemmReshapeRhsMatrixKernel is it possible to set mult_transpose1xW_width, the multiplication factor for the width of the 1xW transposed block
  *
  */
diff --git a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
index f19aa80..bb4c456 100644
--- a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
+++ b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
@@ -111,7 +111,7 @@
      *                              Data types supported: Same as @p input.
      * @param[in]  conv_info        Contains padding and stride information described in @ref PadStrideInfo.
      * @param[in]  weights_info     Specifies if the weights tensor has been reshaped with NEWeightsReshapeKernel. If this is not part of the fully connected layer the weights
-     *                              tensor has also been transposed with NEGEMMTranspose1xWKernel. Data type supported: Same as @p input.
+     *                              tensor has also been transposed with cpu::kernels::CpuGemmTranspose1xWKernel. Data type supported: Same as @p input.
      * @param[in]  dilation         (Optional) Dilation, in elements, across x and y. Defaults to (1, 1).
      * @param[in]  act_info         (Optional) Activation layer information in case of a fused activation. Only RELU, BOUNDED_RELU and LU_BOUNDED_RELU supported.
      * @param[in]  enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation
@@ -133,7 +133,7 @@
      *                             Data types supported: Same as @p input.
      * @param[in] conv_info        Contains padding and stride information described in @ref PadStrideInfo.
      * @param[in] weights_info     Specifies if the weights tensor has been reshaped with NEWeightsReshapeKernel. If this is not part of the fully connected layer the weights
-     *                             tensor has also been transposed with NEGEMMTranspose1xWKernel. Data type supported: Same as @p input.
+     *                             tensor has also been transposed with cpu::kernels::CpuGemmTranspose1xWKernel. Data type supported: Same as @p input.
      * @param[in] dilation         (Optional) Dilation, in elements, across x and y. Defaults to (1, 1).
      * @param[in] act_info         (Optional) Activation layer information in case of a fused activation.
      * @param[in] enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation
@@ -156,7 +156,7 @@
      *                             Data types supported: Same as @p input.
      * @param[in] conv_info        Contains padding and stride information described in @ref PadStrideInfo.
      * @param[in] weights_info     Specifies if the weights tensor has been reshaped with NEWeightsReshapeKernel. If this is not part of the fully connected layer the weights
-     *                             tensor has also been transposed with NEGEMMTranspose1xWKernel. Data type supported: Same as @p input.
+     *                             tensor has also been transposed with cpu::kernels::CpuGemmTranspose1xWKernel. Data type supported: Same as @p input.
      * @param[in] dilation         (Optional) Dilation, in elements, across x and y. Defaults to (1, 1).
      * @param[in] act_info         (Optional) Activation layer information in case of a fused activation.
      * @param[in] enable_fast_math (Optional) Enable fast math computation. In case this flag were set, the function could dispatch the fastest implementation
diff --git a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h
index d2cd60e..22ec9e0 100644
--- a/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h
+++ b/arm_compute/runtime/NEON/functions/NEFullyConnectedLayer.h
@@ -80,7 +80,7 @@
  *  -# @ref NEIm2ColKernel (called when the input comes from a convolutional layer)
  *  -# @ref NETranspose (if @p are_weights_reshaped is set to false and transpose_weights is set to true ) (called once)
  *  -# @ref NEGEMMMatrixMultiplyKernel or @ref NEGEMMLowpMatrixMultiplyCore (if quantized asymmetric)
- *  -# @ref NEGEMMMatrixAdditionKernel or @ref NEGEMMLowpOutputStage (if quantized asymmetric) (if @p biases is not equal to nullptr)
+ *  -# @ref cpu::kernels::CpuGemmMatrixAdditionKernel or @ref NEGEMMLowpOutputStage (if quantized asymmetric) (if @p biases is not equal to nullptr)
  *
  * @note  The fully connected layer accepts "weights" tensors only with 2 dimensions.
  */
diff --git a/arm_compute/runtime/NEON/functions/NEGEMM.h b/arm_compute/runtime/NEON/functions/NEGEMM.h
index 6c5be0e..c1ae11b 100644
--- a/arm_compute/runtime/NEON/functions/NEGEMM.h
+++ b/arm_compute/runtime/NEON/functions/NEGEMM.h
@@ -29,40 +29,26 @@
 #include "arm_compute/runtime/IMemoryManager.h"
 #include "arm_compute/runtime/IWeightsManager.h"
 #include "arm_compute/runtime/MemoryGroup.h"
-#include "arm_compute/runtime/NEON/functions/NEActivationLayer.h"
-#include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h"
-#include "arm_compute/runtime/Tensor.h"
-#include "src/core/helpers/MemoryHelpers.h"
 
 #include <memory>
 
 namespace arm_compute
 {
-// Forward declarations
-class NEGEMMInterleave4x4Kernel;
-class NEGEMMMatrixAdditionKernel;
-class NEGEMMMatrixMultiplyKernel;
-class NEGEMMTranspose1xWKernel;
-namespace cpu
-{
-class CpuGemmAssemblyDispatch;
-}
-
 /** Basic function to execute GEMM. This function calls the following kernels:
  *
  * If optimized assembly is available:
  *  -# @ref cpu::CpuGemmAssemblyDispatch
- *  -# @ref NEActivationLayer (if alpha != 1.0)
+ *  -# @ref cpu::CpuActivation (if alpha != 1.0)
  * Else:
- *  -# @ref NEGEMMInterleave4x4Kernel (if the output tensor is a matrix)
- *  -# @ref NEGEMMTranspose1xWKernel (if the output tensor is a matrix)
+ *  -# @ref cpu::kernels::CpuGemmInterleave4x4Kernel (if the output tensor is a matrix)
+ *  -# @ref cpu::kernels::CpuGemmTranspose1xWKernel (if the output tensor is a matrix)
  *  -# @ref NEGEMMMatrixMultiplyKernel
  * In both cases:
- *  -# @ref NEGEMMMatrixAdditionKernel (if c != nullptr and beta != 0.0 and is not reshaped once)
+ *  -# @ref cpu::kernels::CpuGemmMatrixAdditionKernel (if c != nullptr and beta != 0.0 and is not reshaped once)
  * Else:
- *  -# @ref NEArithmeticAddition (if c != nullptr and is reshaped once and not optimized assembly in place)
+ *  -# @ref cpu::CpuAdd (if c != nullptr and is reshaped once and not optimized assembly in place)
  *
- *  -# @ref NEActivationLayer (if activation is specified in GEMMInfo)
+ *  -# @ref cpu::CpuActivation (if activation is specified in GEMMInfo)
  */
 class NEGEMM : public IFunction
 {
@@ -117,33 +103,8 @@
     void prepare() override;
 
 private:
-    MemoryGroup                                   _memory_group;
-    IWeightsManager                              *_weights_manager;
-    std::unique_ptr<NEGEMMInterleave4x4Kernel>    _interleave_kernel;
-    std::unique_ptr<NEGEMMTranspose1xWKernel>     _transpose_kernel;
-    std::unique_ptr<NEGEMMMatrixMultiplyKernel>   _mm_kernel;
-    std::unique_ptr<cpu::CpuGemmAssemblyDispatch> _asm_glue;
-    std::unique_ptr<NEGEMMMatrixAdditionKernel>   _ma_kernel;
-    NEActivationLayer                             _alpha_scale_func;
-    NEArithmeticAddition                          _add_bias;
-    NEActivationLayer                             _activation_func;
-
-    Tensor         _tmp_a;
-    Tensor         _tmp_b;
-    Tensor         _tmp_d;
-    const ITensor *_original_b;
-    bool           _run_vector_matrix_multiplication;
-    bool           _run_alpha_scale;
-    bool           _run_addition;
-    bool           _run_bias_addition;
-    bool           _run_activation;
-    bool           _reshape_b_only_on_first_run;
-    bool           _is_prepared;
-
-    ITensorPack                      _asm_glue_run_pack;
-    ITensorPack                      _asm_glue_prep_pack;
-    WorkspaceData<Tensor>            _asm_glue_workspace;
-    experimental::MemoryRequirements _aux_mem_req;
+    struct Impl;
+    std::unique_ptr<Impl> _impl;
 };
 } // namespace arm_compute
 #endif /*ARM_COMPUTE_NEGEMM_H */
diff --git a/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h
index edb58e9..d334d51 100644
--- a/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h
+++ b/arm_compute/runtime/NEON/functions/NEGEMMConvolutionLayer.h
@@ -203,7 +203,7 @@
      *                          Data types supported: Same as @p input.
      * @param[in]  conv_info    Contains padding and stride information described in @ref PadStrideInfo.
      * @param[in]  weights_info Specifies if the weights tensor has been reshaped with NEWeightsReshapeKernel. If this is not part of the fully connected layer the weights
-     *                          tensor has also been transposed with NEGEMMTranspose1xWKernel. Data type supported: Same as @p input.
+     *                          tensor has also been transposed with cpu::kernels::CpuGemmTranspose1xWKernel. Data type supported: Same as @p input.
      * @param[in]  dilation     (Optional) Dilation, in elements, across x and y. Defaults to (1, 1).
      * @param[in]  act_info     (Optional) Activation layer information in case of a fused activation. Only RELU, BOUNDED_RELU and LU_BOUNDED_RELU supported.
      * @param[in]  num_groups   (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is not supported
@@ -223,7 +223,7 @@
      *                         Data types supported: Same as @p input.
      * @param[in] conv_info    Contains padding and stride information described in @ref PadStrideInfo.
      * @param[in] weights_info Specifies if the weights tensor has been reshaped with NEWeightsReshapeKernel. If this is not part of the fully connected layer the weights
-     *                         tensor has also been transposed with NEGEMMTranspose1xWKernel. Data type supported: Same as @p input.
+     *                         tensor has also been transposed with cpu::kernels::CpuGemmTranspose1xWKernel. Data type supported: Same as @p input.
      * @param[in] dilation     (Optional) Dilation, in elements, across x and y. Defaults to (1, 1).
      * @param[in] act_info     (Optional) Activation layer information in case of a fused activation. Only RELU, BOUNDED_RELU and LU_BOUNDED_RELU supported.
      * @param[in] num_groups   (Optional) Number of groups when performing a grouped convolution. num_groups != 1 is not supported
diff --git a/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h b/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h
index ff88876..60cfd8f 100644
--- a/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h
+++ b/arm_compute/runtime/NEON/functions/NEGEMMLowpMatrixMultiplyCore.h
@@ -35,8 +35,8 @@
 class ITensor;
 /** Basic function to execute GEMMLowpMatrixMultiplyCore. This function calls the following kernels if the DOT product instruction is not available:
  *
- *  -# @ref NEGEMMInterleave4x4Kernel
- *  -# @ref NEGEMMTranspose1xWKernel
+ *  -# @ref cpu::kernels::CpuGemmInterleave4x4Kernel
+ *  -# @ref cpu::kernels::CpuGemmTranspose1xWKernel
  *  -# @ref NEGEMMLowpMatrixMultiplyKernel
  *  -# @ref NEGEMMLowpOffsetContributionKernel
  *  -# @ref NEActivationLayer
diff --git a/arm_compute/runtime/NEON/functions/NELSTMLayer.h b/arm_compute/runtime/NEON/functions/NELSTMLayer.h
index 075fb45..4272215 100644
--- a/arm_compute/runtime/NEON/functions/NELSTMLayer.h
+++ b/arm_compute/runtime/NEON/functions/NELSTMLayer.h
@@ -25,7 +25,7 @@
 #define ARM_COMPUTE_NELSTMLAYER_H
 
 #include "arm_compute/core/Types.h"
-#include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h"
+#include "arm_compute/runtime/NEON/functions/NEActivationLayer.h"
 #include "arm_compute/runtime/NEON/functions/NEArithmeticAddition.h"
 #include "arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h"
 #include "arm_compute/runtime/NEON/functions/NEConcatenateLayer.h"