diff --git a/tests/validation/Helpers.cpp b/tests/validation/Helpers.cpp
index 7ff2ab6..b53d46f 100644
--- a/tests/validation/Helpers.cpp
+++ b/tests/validation/Helpers.cpp
@@ -325,6 +325,30 @@
     return std::pair<int, int> { min_bound, max_bound };
 }
 
+void add_padding_x(std::initializer_list<ITensor *> tensors, const DataLayout &data_layout)
+{
+    if(data_layout == DataLayout::NHWC)
+    {
+        constexpr unsigned int lower = 1U;
+        constexpr unsigned int upper = 16U;
+
+        std::uniform_int_distribution<unsigned int> distribution(lower, upper);
+        size_t                                      seed_offset = 0;
+
+        for(ITensor *tensor : tensors)
+        {
+            ARM_COMPUTE_ERROR_ON(!tensor->info()->is_resizable());
+
+            std::mt19937 gen(library->seed() + seed_offset++);
+
+            const unsigned int right = distribution(gen);
+            const unsigned int left  = distribution(gen);
+
+            tensor->info()->extend_padding(PaddingSize(0U, right, 0U, left));
+        }
+    }
+}
+
 template void get_tile(const SimpleTensor<float> &in, SimpleTensor<float> &roi, const Coordinates &coord);
 template void get_tile(const SimpleTensor<half> &in, SimpleTensor<half> &roi, const Coordinates &coord);
 template void get_tile(const SimpleTensor<int> &in, SimpleTensor<int> &roi, const Coordinates &coord);
diff --git a/tests/validation/Helpers.h b/tests/validation/Helpers.h
index 30ec14e..3ba3bd1 100644
--- a/tests/validation/Helpers.h
+++ b/tests/validation/Helpers.h
@@ -229,6 +229,15 @@
  * @param[in] channel_id Channel id for per channel quantization info.
  */
 std::pair<int, int> get_symm_quantized_per_channel_bounds(const QuantizationInfo &quant_info, float min, float max, size_t channel_id = 0);
+
+/** Add random padding along the X axis (between 1 and 16 columns per side) to all the input tensors
+ *
+ * @param[in] tensors     List of tensors to add padding to
+ * @param[in] data_layout (Optional) Data layout of the operator
+ *
+ * @note This function adds padding to the input tensors only if data_layout == DataLayout::NHWC
+ */
+void add_padding_x(std::initializer_list<ITensor *> tensors, const DataLayout &data_layout = DataLayout::NHWC);
 } // namespace validation
 } // namespace test
 } // namespace arm_compute
diff --git a/tests/validation/NEON/GEMMLowp.cpp b/tests/validation/NEON/GEMMLowp.cpp
index 518f480..9d075e1 100644
--- a/tests/validation/NEON/GEMMLowp.cpp
+++ b/tests/validation/NEON/GEMMLowp.cpp
@@ -37,7 +37,6 @@
 #include "tests/framework/Macros.h"
 #include "tests/framework/datasets/Datasets.h"
 #include "tests/validation/Validation.h"
-#include "tests/validation/fixtures/GEMMLowpAssemblyFixture.h"
 #include "tests/validation/fixtures/GEMMLowpFixture.h"
 
 namespace arm_compute
diff --git a/tests/validation/UNIT/GPUTarget.cpp b/tests/validation/UNIT/GPUTarget.cpp
index e1b7e1f..d2c81cf 100644
--- a/tests/validation/UNIT/GPUTarget.cpp
+++ b/tests/validation/UNIT/GPUTarget.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018-2020 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -22,7 +22,6 @@
  * SOFTWARE.
  */
 #include "arm_compute/core/GPUTarget.h"
-#include "tests/AssetsLibrary.h"
 #include "tests/Globals.h"
 #include "tests/Utils.h"
 #include "tests/framework/Asserts.h"
diff --git a/tests/validation/UNIT/SafeIntegerOps.cpp b/tests/validation/UNIT/SafeIntegerOps.cpp
index 62f7041..13e4ef5 100644
--- a/tests/validation/UNIT/SafeIntegerOps.cpp
+++ b/tests/validation/UNIT/SafeIntegerOps.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2019-2020 Arm Limited.
+ * Copyright (c) 2019-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -23,7 +23,6 @@
  */
 #include "arm_compute/core/GPUTarget.h"
 #include "arm_compute/core/utils/math/SafeOps.h"
-#include "tests/AssetsLibrary.h"
 #include "tests/Globals.h"
 #include "tests/Utils.h"
 #include "tests/framework/Asserts.h"
diff --git a/tests/validation/fixtures/ConvolutionLayerFixture.h b/tests/validation/fixtures/ConvolutionLayerFixture.h
index 07790e8..b649280 100644
--- a/tests/validation/fixtures/ConvolutionLayerFixture.h
+++ b/tests/validation/fixtures/ConvolutionLayerFixture.h
@@ -69,7 +69,8 @@
 public:
     template <typename...>
     void setup(TensorShape input_shape, TensorShape weights_shape, TensorShape bias_shape, TensorShape output_shape, PadStrideInfo info, Size2D dilation, bool reshape_weights,
-               DataType data_type, DataType weights_data_type, DataLayout data_layout, QuantizationInfo quantization_info, QuantizationInfo weight_quantization_info, ActivationLayerInfo act_info, bool mixed_layout = false)
+               DataType data_type, DataType weights_data_type, DataLayout data_layout, QuantizationInfo quantization_info, QuantizationInfo weight_quantization_info, ActivationLayerInfo act_info,
+               bool mixed_layout = false)
     {
         _mixed_layout             = mixed_layout;
         _data_type                = data_type;
@@ -87,7 +88,6 @@
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         // Test Multi DataLayout graph cases, when the data layout changes after configure
@@ -214,6 +214,8 @@
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &weights, &bias, &dst }, _data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         weights.allocator()->allocate();
diff --git a/tests/validation/fixtures/DepthwiseConvolutionLayerFixture.h b/tests/validation/fixtures/DepthwiseConvolutionLayerFixture.h
index 0aa43d8..e87e31f 100644
--- a/tests/validation/fixtures/DepthwiseConvolutionLayerFixture.h
+++ b/tests/validation/fixtures/DepthwiseConvolutionLayerFixture.h
@@ -116,6 +116,9 @@
 
     void allocate_and_run_target()
     {
+        // TODO: uncomment after COMPMID-4361
+        // add_padding_x({ &_src, &_weights, &_biases, &_target }, _data_layout);
+
         // Allocate tensors
         _src.allocator()->allocate();
         _weights.allocator()->allocate();
@@ -131,7 +134,7 @@
         fill(AccessorType(_src), 0);
         fill(AccessorType(_weights), 1);
         fill(AccessorType(_biases), 2);
-        
+
         if(_mixed_layout)
         {
             mix_layout(_dwc, _src, _target);
@@ -158,7 +161,6 @@
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         // Test Multi DataLayout graph cases, when the data layout changes after configure
@@ -237,7 +239,7 @@
     ActivationLayerInfo _act_info{};
     unsigned int        _depth_multiplier{};
     Size2D              _dilation{};
-    bool                _mixed_layout{false};
+    bool                _mixed_layout{ false };
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T, bool mixed_layout = false>
@@ -309,6 +311,8 @@
 
     void allocate_and_run_target()
     {
+        add_padding_x({ &_src, &_weights, &_biases, &_target }, _data_layout);
+
         // Allocate tensors
         _src.allocator()->allocate();
         _weights.allocator()->allocate();
@@ -442,6 +446,8 @@
 
     void allocate_and_run_target()
     {
+        add_padding_x({ &_src, &_weights, &_biases, &_target }, _data_layout);
+
         // Allocate tensors
         _src.allocator()->allocate();
         _weights.allocator()->allocate();
diff --git a/tests/validation/fixtures/DirectConvolutionLayerFixture.h b/tests/validation/fixtures/DirectConvolutionLayerFixture.h
index 5ed0b9f..b79991e 100644
--- a/tests/validation/fixtures/DirectConvolutionLayerFixture.h
+++ b/tests/validation/fixtures/DirectConvolutionLayerFixture.h
@@ -90,7 +90,6 @@
     }
 
 protected:
-    
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         DataLayout data_layout = src.info()->data_layout();
@@ -172,6 +171,9 @@
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: uncomment after COMPMID-4341
+        // add_padding_x({ &src, &weights, &bias, &dst }, data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         weights.allocator()->allocate();
@@ -221,7 +223,7 @@
     SimpleTensor<T>  _reference{};
     QuantizationInfo _quantization_info{};
     DataType         _data_type{};
-    bool             _mixed_layout {false};
+    bool             _mixed_layout{ false };
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T, bool mixed_layout = false>
diff --git a/tests/validation/fixtures/FFTFixture.h b/tests/validation/fixtures/FFTFixture.h
index 199730d..3a75135 100644
--- a/tests/validation/fixtures/FFTFixture.h
+++ b/tests/validation/fixtures/FFTFixture.h
@@ -91,6 +91,9 @@
         ARM_COMPUTE_EXPECT(src.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: uncomment after COMPMID-4362
+        // add_padding_x({ &src, &dst });
+
         // Allocate tensors
         src.allocator()->allocate();
         dst.allocator()->allocate();
@@ -137,15 +140,14 @@
                DataType data_type, DataLayout data_layout, ActivationLayerInfo act_info, bool mixed_layout = false)
     {
         _mixed_layout = mixed_layout;
-        _data_type   = data_type;
-        _data_layout = data_layout;
+        _data_type    = data_type;
+        _data_layout  = data_layout;
 
         _target    = compute_target(input_shape, weights_shape, bias_shape, output_shape, info, dilation, act_info);
         _reference = compute_reference(input_shape, weights_shape, bias_shape, output_shape, info, dilation, act_info);
     }
 
 protected:
-    
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         // Test Multi DataLayout graph cases, when the data layout changes after configure
@@ -210,6 +212,8 @@
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &weights, &bias, &dst }, _data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         weights.allocator()->allocate();
@@ -225,7 +229,7 @@
         fill(AccessorType(src), 0);
         fill(AccessorType(weights), 1);
         fill(AccessorType(bias), 2);
-        
+
         if(_mixed_layout)
         {
             mix_layout(conv, src, dst);
@@ -261,7 +265,7 @@
     SimpleTensor<T> _reference{};
     DataType        _data_type{};
     DataLayout      _data_layout{};
-    bool            _mixed_layout{false};
+    bool            _mixed_layout{ false };
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T, bool mixed_layout = false>
diff --git a/tests/validation/fixtures/FullyConnectedLayerFixture.h b/tests/validation/fixtures/FullyConnectedLayerFixture.h
index 8f38aae..383d880 100644
--- a/tests/validation/fixtures/FullyConnectedLayerFixture.h
+++ b/tests/validation/fixtures/FullyConnectedLayerFixture.h
@@ -72,7 +72,6 @@
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         const DataLayout data_layout = src.info()->data_layout();
@@ -165,6 +164,8 @@
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &weights, &bias, &dst });
+
         // Allocate tensors
         src.allocator()->allocate();
         weights.allocator()->allocate();
@@ -238,7 +239,7 @@
     SimpleTensor<T>     _reference{};
     DataType            _data_type{};
     DataType            _bias_data_type{};
-    bool                _mixed_layout{false};
+    bool                _mixed_layout{ false };
     QuantizationInfo    _quantization_info{};
     ActivationLayerInfo _activation_info{};
 };
diff --git a/tests/validation/fixtures/GEMMFixture.h b/tests/validation/fixtures/GEMMFixture.h
index 500e094..45516d4 100644
--- a/tests/validation/fixtures/GEMMFixture.h
+++ b/tests/validation/fixtures/GEMMFixture.h
@@ -105,6 +105,8 @@
         ARM_COMPUTE_EXPECT(c.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &a, &b, &c, &dst });
+
         // Allocate tensors
         a.allocator()->allocate();
         b.allocator()->allocate();
@@ -231,6 +233,8 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &bias, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -347,6 +351,8 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &bias, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -483,6 +489,12 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: remove if statement after COMPMID-4368
+        if(!rhs_info.export_to_cl_image)
+        {
+            add_padding_x({ &lhs, &rhs, &lhs_reshaped, &rhs_reshaped, &bias, &dst });
+        }
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -625,6 +637,12 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: remove if statement after COMPMID-4368
+        if(!rhs_info.export_to_cl_image)
+        {
+            add_padding_x({ &lhs, &rhs, &lhs_reshaped, &rhs_reshaped, &bias, &dst });
+        }
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -787,6 +805,12 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: remove if statement after COMPMID-4368
+        if(!rhs_info.export_to_cl_image)
+        {
+            add_padding_x({ &lhs, &rhs, &lhs_reshaped, &rhs_reshaped, &bias, &dst });
+        }
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -955,6 +979,12 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: remove if statement after COMPMID-4368
+        if(!rhs_info.export_to_cl_image)
+        {
+            add_padding_x({ &lhs, &rhs, &lhs_reshaped, &rhs_reshaped, &bias, &dst });
+        }
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1118,6 +1148,12 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: remove if statement after COMPMID-4368
+        if(!rhs_info.export_to_cl_image)
+        {
+            add_padding_x({ &lhs, &rhs, &rhs_reshaped, &bias, &dst });
+        }
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1277,6 +1313,12 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: remove if statement after COMPMID-4368
+        if(!rhs_info.export_to_cl_image)
+        {
+            add_padding_x({ &lhs, &rhs, &rhs_reshaped, &bias, &dst });
+        }
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1410,6 +1452,8 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &bias, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1539,6 +1583,8 @@
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &bias, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
diff --git a/tests/validation/fixtures/GEMMLowpAssemblyFixture.h b/tests/validation/fixtures/GEMMLowpAssemblyFixture.h
deleted file mode 100644
index e9ec1bc..0000000
--- a/tests/validation/fixtures/GEMMLowpAssemblyFixture.h
+++ /dev/null
@@ -1,141 +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_TEST_GEMMLOWP_ASSEMBLY_FIXTURE
-#define ARM_COMPUTE_TEST_GEMMLOWP_ASSEMBLY_FIXTURE
-
-#include "arm_compute/core/TensorShape.h"
-#include "arm_compute/core/Types.h"
-#include "tests/AssetsLibrary.h"
-#include "tests/Globals.h"
-#include "tests/IAccessor.h"
-#include "tests/framework/Asserts.h"
-#include "tests/framework/Fixture.h"
-#include "tests/validation/Helpers.h"
-#include "tests/validation/reference/GEMMLowp.h"
-
-#include <random>
-
-namespace arm_compute
-{
-namespace test
-{
-namespace validation
-{
-template <typename TensorType, typename AccessorType, typename FunctionType, typename T2>
-class GEMMLowpAssemblyFixture : public framework::Fixture
-{
-public:
-    template <typename...>
-    void setup(size_t m, size_t n, size_t k)
-    {
-        const TensorShape shape_a(k, m);
-        const TensorShape shape_b(n, k);
-        const TensorShape shape_c(n, m);
-        _target    = compute_target(shape_a, shape_b, shape_c);
-        _reference = compute_reference(shape_a, shape_b, shape_c);
-    }
-
-protected:
-    template <typename U>
-    void fill(U &&tensor, int i, int lo, int hi)
-    {
-        std::uniform_int_distribution<> distribution(lo, hi);
-        library->fill(tensor, distribution, i);
-    }
-
-    TensorType compute_target(const TensorShape &shape_a, const TensorShape &shape_b, const TensorShape &shape_c)
-    {
-        DataType dt_in = std::is_same<T2, int8_t>::value ? DataType::S8 : DataType::U8;
-
-        // Create tensors
-        TensorType a = create_tensor<TensorType>(shape_a, dt_in, 1);
-        TensorType b = create_tensor<TensorType>(shape_b, dt_in, 1);
-        TensorType c = create_tensor<TensorType>(shape_c, DataType::S32, 1);
-
-        // Create and configure function
-        FunctionType gemmlowp;
-        gemmlowp.configure(&a, &b, nullptr, &c);
-
-        ARM_COMPUTE_EXPECT(a.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(b.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(c.info()->is_resizable(), framework::LogLevel::ERRORS);
-
-        // Allocate tensors
-        a.allocator()->allocate();
-        b.allocator()->allocate();
-        c.allocator()->allocate();
-
-        ARM_COMPUTE_EXPECT(!a.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(!b.info()->is_resizable(), framework::LogLevel::ERRORS);
-        ARM_COMPUTE_EXPECT(!c.info()->is_resizable(), framework::LogLevel::ERRORS);
-
-        // Fill tensors
-        if(dt_in == DataType::S8)
-        {
-            fill(AccessorType(a), 0, -128, 127);
-            fill(AccessorType(b), 1, -128, 127);
-        }
-        else
-        {
-            fill(AccessorType(a), 0, 0, 255);
-            fill(AccessorType(b), 1, 0, 255);
-        }
-        fill(AccessorType(c), 2, 0, 0);
-
-        // Compute GEMM function
-        gemmlowp.run();
-        return c;
-    }
-
-    SimpleTensor<int32_t> compute_reference(const TensorShape &shape_a, const TensorShape &shape_b, const TensorShape &shape_c)
-    {
-        DataType dt = std::is_same<T2, int8_t>::value ? DataType::S8 : DataType::U8;
-
-        // Create reference
-        SimpleTensor<T2> a{ shape_a, dt, 1 };
-        SimpleTensor<T2> b{ shape_b, dt, 1 };
-
-        // Fill reference
-        if(dt == DataType::S8)
-        {
-            fill(a, 0, -128, 127);
-            fill(b, 1, -128, 127);
-        }
-        else
-        {
-            fill(a, 0, 0, 255);
-            fill(b, 1, 0, 255);
-        }
-
-        return reference::gemmlowp<int32_t, T2>(a, b, shape_c);
-    }
-
-    TensorType            _target{};
-    SimpleTensor<int32_t> _reference{};
-};
-
-} // namespace validation
-} // namespace test
-} // namespace arm_compute
-#endif /* ARM_COMPUTE_TEST_GEMMLOWP_FIXTURE */
diff --git a/tests/validation/fixtures/GEMMLowpFixture.h b/tests/validation/fixtures/GEMMLowpFixture.h
index 95f4960..c3da2e2 100644
--- a/tests/validation/fixtures/GEMMLowpFixture.h
+++ b/tests/validation/fixtures/GEMMLowpFixture.h
@@ -133,6 +133,8 @@
     ARM_COMPUTE_EXPECT(b.info()->is_resizable(), framework::LogLevel::ERRORS);
     ARM_COMPUTE_EXPECT(output.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+    add_padding_x({ &a, &b, &output });
+
     // Allocate tensors
     a.allocator()->allocate();
     b.allocator()->allocate();
@@ -948,6 +950,8 @@
         ARM_COMPUTE_EXPECT(lhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &lhs_reshaped, &rhs_reshaped, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1098,6 +1102,8 @@
         ARM_COMPUTE_EXPECT(lhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &lhs_reshaped, &rhs_reshaped, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1247,6 +1253,8 @@
         ARM_COMPUTE_EXPECT(lhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &rhs_reshaped, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1390,6 +1398,8 @@
         ARM_COMPUTE_EXPECT(lhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &rhs_reshaped, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1502,6 +1512,8 @@
         ARM_COMPUTE_EXPECT(lhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
@@ -1597,6 +1609,8 @@
         ARM_COMPUTE_EXPECT(lhs.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(rhs.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &lhs, &rhs, &dst });
+
         // Allocate tensors
         lhs.allocator()->allocate();
         rhs.allocator()->allocate();
diff --git a/tests/validation/fixtures/GEMMReshapeLHSMatrixFixture.h b/tests/validation/fixtures/GEMMReshapeLHSMatrixFixture.h
index d085509..4ee493b 100644
--- a/tests/validation/fixtures/GEMMReshapeLHSMatrixFixture.h
+++ b/tests/validation/fixtures/GEMMReshapeLHSMatrixFixture.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -91,6 +91,8 @@
 
         ARM_COMPUTE_EXPECT(src.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &dst });
+
         // Allocate tensors
         src.allocator()->allocate();
         dst.allocator()->allocate();
diff --git a/tests/validation/fixtures/GEMMReshapeRHSMatrixFixture.h b/tests/validation/fixtures/GEMMReshapeRHSMatrixFixture.h
index 99bfa3b..3f73912 100644
--- a/tests/validation/fixtures/GEMMReshapeRHSMatrixFixture.h
+++ b/tests/validation/fixtures/GEMMReshapeRHSMatrixFixture.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2018 Arm Limited.
+ * Copyright (c) 2018-2021 Arm Limited.
  *
  * SPDX-License-Identifier: MIT
  *
@@ -90,6 +90,8 @@
 
         ARM_COMPUTE_EXPECT(src.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &dst });
+
         // Allocate tensors
         src.allocator()->allocate();
         dst.allocator()->allocate();
diff --git a/tests/validation/fixtures/PoolingLayerFixture.h b/tests/validation/fixtures/PoolingLayerFixture.h
index ee81ff5..66e09d5 100644
--- a/tests/validation/fixtures/PoolingLayerFixture.h
+++ b/tests/validation/fixtures/PoolingLayerFixture.h
@@ -50,13 +50,12 @@
                QuantizationInfo input_qinfo = QuantizationInfo(), QuantizationInfo output_qinfo = QuantizationInfo(), bool mixed_layout = false)
     {
         _mixed_layout = mixed_layout;
-        _pool_info = pool_info;
-        _target    = compute_target(shape, pool_info, data_type, data_layout, input_qinfo, output_qinfo, indices);
-        _reference = compute_reference(shape, pool_info, data_type, data_layout, input_qinfo, output_qinfo, indices);
+        _pool_info    = pool_info;
+        _target       = compute_target(shape, pool_info, data_type, data_layout, input_qinfo, output_qinfo, indices);
+        _reference    = compute_reference(shape, pool_info, data_type, data_layout, input_qinfo, output_qinfo, indices);
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         const DataLayout data_layout = src.info()->data_layout();
@@ -115,6 +114,9 @@
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(_target_indices.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        // TODO: uncomment after COMPMID-4363
+        // add_padding_x({ &src, &dst, &_target_indices }, data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         dst.allocator()->allocate();
@@ -152,7 +154,7 @@
     TensorType             _target{};
     SimpleTensor<T>        _reference{};
     PoolingLayerInfo       _pool_info{};
-    bool                   _mixed_layout{false};
+    bool                   _mixed_layout{ false };
     TensorType             _target_indices{};
     SimpleTensor<uint32_t> _ref_indices{};
 };
diff --git a/tests/validation/fixtures/ScaleFixture.h b/tests/validation/fixtures/ScaleFixture.h
index 9e0f620..a40cfda 100644
--- a/tests/validation/fixtures/ScaleFixture.h
+++ b/tests/validation/fixtures/ScaleFixture.h
@@ -68,7 +68,6 @@
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         const DataLayout data_layout = src.info()->data_layout();
@@ -163,6 +162,8 @@
         ARM_COMPUTE_EXPECT(src.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &dst }, data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         dst.allocator()->allocate();
diff --git a/tests/validation/fixtures/WinogradConvolutionLayerFixture.h b/tests/validation/fixtures/WinogradConvolutionLayerFixture.h
index f956963..a1433e9 100644
--- a/tests/validation/fixtures/WinogradConvolutionLayerFixture.h
+++ b/tests/validation/fixtures/WinogradConvolutionLayerFixture.h
@@ -62,12 +62,11 @@
     {
         ARM_COMPUTE_UNUSED(dilation);
         _mixed_layout = mixed_layout;
-        _target    = compute_target(input_shape, weights_shape, bias_shape, output_shape, info, data_type, act_info, data_layout);
-        _reference = compute_reference(input_shape, weights_shape, bias_shape, info, data_type, act_info);
+        _target       = compute_target(input_shape, weights_shape, bias_shape, output_shape, info, data_type, act_info, data_layout);
+        _reference    = compute_reference(input_shape, weights_shape, bias_shape, info, data_type, act_info);
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         const DataLayout data_layout = src.info()->data_layout();
@@ -134,6 +133,8 @@
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &weights, &bias, &dst }, data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         weights.allocator()->allocate();
@@ -235,7 +236,7 @@
 
     TensorType      _target{};
     SimpleTensor<T> _reference{};
-    bool            _mixed_layout{false};
+    bool            _mixed_layout{ false };
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T, bool mixed_layout = false>
@@ -246,13 +247,12 @@
     void setup(TensorShape input_shape, WinogradInfo winograd_info, DataLayout data_layout, DataType data_type)
     {
         TensorShape output_shape = compute_winograd_input_transform_shape(TensorInfo(input_shape, 1, data_type), winograd_info);
-        _mixed_layout = mixed_layout;
-        _target    = compute_target(input_shape, output_shape, winograd_info, data_layout, data_type);
-        _reference = compute_reference(input_shape, output_shape, winograd_info, data_type);
+        _mixed_layout            = mixed_layout;
+        _target                  = compute_target(input_shape, output_shape, winograd_info, data_layout, data_type);
+        _reference               = compute_reference(input_shape, output_shape, winograd_info, data_type);
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         const DataLayout data_layout_src = src.info()->data_layout();
@@ -311,6 +311,8 @@
         ARM_COMPUTE_EXPECT(src.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &dst }, data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         dst.allocator()->allocate();
@@ -344,7 +346,7 @@
         return reference::winograd_input_transform<T>(src, output_shape, winograd_info);
     }
 
-    bool _mixed_layout {false};
+    bool            _mixed_layout{ false };
     TensorType      _target{};
     SimpleTensor<T> _reference{};
 };
@@ -360,12 +362,11 @@
         TensorShape  output_shape = compute_winograd_filter_transform_shape(TensorInfo(input_shape, 1, data_type), winograd_info);
 
         _mixed_layout = mixed_layout;
-        _target    = compute_target(input_shape, output_shape, winograd_info, data_layout, data_type);
-        _reference = compute_reference(input_shape, output_shape, winograd_info, data_type);
+        _target       = compute_target(input_shape, output_shape, winograd_info, data_layout, data_type);
+        _reference    = compute_reference(input_shape, output_shape, winograd_info, data_type);
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         const DataLayout data_layout_src = src.info()->data_layout();
@@ -425,6 +426,8 @@
         ARM_COMPUTE_EXPECT(src.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &dst }, data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         dst.allocator()->allocate();
@@ -458,7 +461,7 @@
         return reference::winograd_filter_transform<T>(src, output_shape, winograd_info);
     }
 
-    bool    _mixed_layout {false};
+    bool            _mixed_layout{ false };
     TensorType      _target{};
     SimpleTensor<T> _reference{};
 };
@@ -475,7 +478,6 @@
     }
 
 protected:
-
     void mix_layout(FunctionType &layer, TensorType &src, TensorType &dst)
     {
         const DataLayout data_layout_src = src.info()->data_layout();
@@ -534,6 +536,8 @@
         ARM_COMPUTE_EXPECT(bias.info()->is_resizable(), framework::LogLevel::ERRORS);
         ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
 
+        add_padding_x({ &src, &bias, &dst }, winograd_info.output_data_layout);
+
         // Allocate tensors
         src.allocator()->allocate();
         bias.allocator()->allocate();
@@ -577,7 +581,7 @@
         return (act_info.enabled()) ? reference::activation_layer<T>(winograd_output, act_info) : winograd_output;
     }
 
-    bool    _mixed_layout {false};
+    bool            _mixed_layout{ false };
     TensorType      _target{};
     SimpleTensor<T> _reference{};
 };