diff --git a/arm_compute/runtime/IScheduler.h b/arm_compute/runtime/IScheduler.h
index 6078abd..8918843 100644
--- a/arm_compute/runtime/IScheduler.h
+++ b/arm_compute/runtime/IScheduler.h
@@ -35,23 +35,23 @@
 {
 public:
     /** Default constructor. */
-    IScheduler()
-        : _target(CPUTarget::INTRINSICS)
-    {
-    }
+    IScheduler();
 
     /** Destructor. */
     virtual ~IScheduler() = default;
+
     /** Sets the number of threads the scheduler will use to run the kernels.
      *
      * @param[in] num_threads If set to 0, then one thread per CPU core available on the system will be used, otherwise the number of threads specified.
      */
     virtual void set_num_threads(unsigned int num_threads) = 0;
+
     /** Returns the number of threads that the SingleThreadScheduler has in his pool.
      *
      * @return Number of threads available in SingleThreadScheduler.
      */
     virtual unsigned int num_threads() const = 0;
+
     /** Runs the kernel in the same thread as the caller synchronously.
      *
      * @param[in] kernel          Kernel to execute.
@@ -65,24 +65,14 @@
      */
     void set_target(CPUTarget target);
 
-    /** Return the current CPU target.
+    /** Get CPU info.
      *
-     * @return Target CPU.
+     * @return CPU info.
      */
-    CPUTarget target() const;
+    CPUInfo cpu_info() const;
 
 protected:
-    CPUTarget _target;
+    CPUInfo _info{};
 };
-
-inline void IScheduler::set_target(CPUTarget target)
-{
-    _target = target;
-}
-
-inline CPUTarget IScheduler::target() const
-{
-    return _target;
-}
 }
 #endif /* __ARM_COMPUTE_ISCHEDULER_H__ */
diff --git a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
index 8e040b3..893dfa0 100644
--- a/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
+++ b/arm_compute/runtime/NEON/functions/NEConvolutionLayer.h
@@ -28,6 +28,7 @@
 
 #include "arm_compute/core/NEON/kernels/NECol2ImKernel.h"
 #include "arm_compute/core/NEON/kernels/NEFillBorderKernel.h"
+#include "arm_compute/core/NEON/kernels/NEGEMMAssemblyBaseKernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMInterleave4x4Kernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMTranspose1xWKernel.h"
@@ -37,6 +38,8 @@
 #include "arm_compute/runtime/MemoryGroup.h"
 #include "arm_compute/runtime/Tensor.h"
 
+#include <memory>
+
 namespace arm_compute
 {
 class ITensor;
@@ -59,6 +62,7 @@
      *                          Data types supported: Same as @p weights.
      */
     void configure(const ITensor *weights, const ITensor *biases, ITensor *output, bool transpose1xW);
+
     // Inherited methods overridden:
     void run() override;
 
@@ -82,6 +86,7 @@
 public:
     /** Constructor */
     NEConvolutionLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr);
+
     /** Set the input and output tensors.
      *
      * @param[in]  input        Source tensor. 3 lower dimensions represent a single input [width, height, IFM],
@@ -96,23 +101,26 @@
      *                          tensor has also been transposed with NEGEMMTranspose1xWKernel. Data type supported: Same as @p input.
      */
     void configure(const ITensor *input, const ITensor *weights, const ITensor *biases, ITensor *output, const PadStrideInfo &conv_info, const WeightsInfo &weights_info = WeightsInfo());
+
     // Inherited methods overridden:
     void run() override;
 
 private:
-    MemoryGroup                      _memory_group;
-    NEIm2ColKernel                   _input_im2col_kernel;
-    NEGEMMInterleave4x4Kernel        _input_interleave_kernel;
-    NEConvolutionLayerReshapeWeights _reshape_weights;
-    NEGEMMMatrixMultiplyKernel       _mm_kernel;
-    NECol2ImKernel                   _output_col2im_kernel;
-    Tensor                           _input_im2col_reshaped;
-    Tensor                           _input_interleaved_reshaped;
-    Tensor                           _weights_reshaped;
-    Tensor                           _gemm_output;
-    bool                             _has_bias;
-    bool                             _is_fully_connected_convolution;
-    bool                             _are_weights_reshaped;
+    MemoryGroup                               _memory_group;
+    NEIm2ColKernel                            _input_im2col_kernel;
+    NEGEMMInterleave4x4Kernel                 _input_interleave_kernel;
+    NEConvolutionLayerReshapeWeights          _reshape_weights;
+    NEGEMMMatrixMultiplyKernel                _mm_kernel;
+    std::unique_ptr<NEGEMMAssemblyBaseKernel> _mm_optimised_kernel;
+    NECol2ImKernel                            _output_col2im_kernel;
+    Tensor                                    _input_im2col_reshaped;
+    Tensor                                    _input_interleaved_reshaped;
+    Tensor                                    _weights_reshaped;
+    Tensor                                    _gemm_output;
+    Tensor                                    _workspace;
+    bool                                      _has_bias;
+    bool                                      _is_fully_connected_convolution;
+    bool                                      _are_weights_reshaped;
 };
 }
 #endif /* __ARM_COMPUTE_NECONVOLUTIONLAYER_H__ */
diff --git a/arm_compute/runtime/NEON/functions/NEGEMM.h b/arm_compute/runtime/NEON/functions/NEGEMM.h
index b4b9e8b..068e7c5 100644
--- a/arm_compute/runtime/NEON/functions/NEGEMM.h
+++ b/arm_compute/runtime/NEON/functions/NEGEMM.h
@@ -25,6 +25,7 @@
 #define __ARM_COMPUTE_NEGEMM_H__
 
 #include "arm_compute/core/NEON/kernels/NEFillBorderKernel.h"
+#include "arm_compute/core/NEON/kernels/NEGEMMAssemblyBaseKernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMInterleave4x4Kernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMMatrixAdditionKernel.h"
 #include "arm_compute/core/NEON/kernels/NEGEMMMatrixMultiplyKernel.h"
@@ -51,6 +52,7 @@
 public:
     /** Constructor */
     NEGEMM(std::shared_ptr<IMemoryManager> memory_manager = nullptr);
+
     /** Initialise the kernel's inputs, output
      *
      * @note GEMM: General Matrix Multiply - [alpha * A * B + beta * C].
@@ -69,15 +71,17 @@
     void run() override;
 
 private:
-    MemoryGroup                _memory_group;
-    NEGEMMInterleave4x4Kernel  _interleave_kernel;
-    NEGEMMTranspose1xWKernel   _transpose_kernel;
-    NEGEMMMatrixMultiplyKernel _mm_kernel;
-    NEGEMMMatrixAdditionKernel _ma_kernel;
-    Tensor                     _tmp_a;
-    Tensor                     _tmp_b;
-    bool                       _run_vector_matrix_multiplication;
-    bool                       _run_addition;
+    MemoryGroup                               _memory_group;
+    NEGEMMInterleave4x4Kernel                 _interleave_kernel;
+    NEGEMMTranspose1xWKernel                  _transpose_kernel;
+    NEGEMMMatrixMultiplyKernel                _mm_kernel;
+    std::unique_ptr<NEGEMMAssemblyBaseKernel> _mm_optimised_kernel;
+    NEGEMMMatrixAdditionKernel                _ma_kernel;
+    Tensor                                    _tmp_a;
+    Tensor                                    _tmp_b;
+    Tensor                                    _workspace;
+    bool                                      _run_vector_matrix_multiplication;
+    bool                                      _run_addition;
 };
 }
 #endif /*__ARM_COMPUTE_NEGEMM_H__ */