diff --git a/examples/graph_googlenet.cpp b/examples/graph_googlenet.cpp
index d64512b..e97f3ac 100644
--- a/examples/graph_googlenet.cpp
+++ b/examples/graph_googlenet.cpp
@@ -54,7 +54,7 @@
         // Set target. 0 (NEON), 1 (OpenCL), 2 (OpenCL with Tuner). By default it is NEON
         const int         target                   = argc > 1 ? std::strtol(argv[1], nullptr, 10) : 0;
         Target            target_hint              = set_target_hint2(target);
-        ConvolutionMethod convolution_hint         = ConvolutionMethod::GEMM;
+        ConvolutionMethod convolution_hint         = (target_hint == Target::CL) ? ConvolutionMethod::WINOGRAD : ConvolutionMethod::GEMM;
         bool              enable_tuning            = (target == 2);
         bool              enable_memory_management = true;
 
diff --git a/examples/graph_inception_v3.cpp b/examples/graph_inception_v3.cpp
index 9bb51ba..73a4450 100644
--- a/examples/graph_inception_v3.cpp
+++ b/examples/graph_inception_v3.cpp
@@ -56,6 +56,8 @@
         bool      enable_tuning            = (target == 2);
         bool      enable_memory_management = true;
 
+        ConvolutionMethod convolution_hint = (target_hint == Target::CL) ? ConvolutionMethod::WINOGRAD : ConvolutionMethod::GEMM;
+
         // Parse arguments
         if(argc < 2)
         {
@@ -90,7 +92,6 @@
 
         graph << target_hint << InputLayer(TensorDescriptor(TensorShape(299U, 299U, 3U, 1U), DataType::F32),
                                            get_input_accessor(image, std::move(preprocessor), false))
-
               << ConvolutionLayer(3U, 3U, 32U,
                                   get_weights_accessor(data_path, "/cnn_data/inceptionv3_model/Conv2d_1a_3x3_weights.npy"),
                                   std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr), PadStrideInfo(2, 2, 0, 0))
@@ -102,7 +103,7 @@
                                                                                              "/cnn_data/inceptionv3_model/Conv2d_1a_3x3_BatchNorm_beta.npy"),
                                          0.001f)
               << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
-
+              << convolution_hint
               << ConvolutionLayer(3U, 3U, 32U,
                                   get_weights_accessor(data_path, "/cnn_data/inceptionv3_model/Conv2d_2a_3x3_weights.npy"),
                                   std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr), PadStrideInfo(1, 1, 0, 0))
diff --git a/examples/graph_inception_v4.cpp b/examples/graph_inception_v4.cpp
index 6f76b5e..88073b7 100644
--- a/examples/graph_inception_v4.cpp
+++ b/examples/graph_inception_v4.cpp
@@ -60,6 +60,8 @@
         bool      enable_tuning            = (target == 2);
         bool      enable_memory_management = true;
 
+        ConvolutionMethod convolution_hint = (target_hint == Target::CL) ? ConvolutionMethod::WINOGRAD : ConvolutionMethod::GEMM;
+
         // Parse arguments
         if(argc < 2)
         {
@@ -104,6 +106,7 @@
                                          get_weights_accessor(data_path, "/cnn_data/inceptionv4_model/Conv2d_1a_3x3_BatchNorm_beta.npy"),
                                          0.001f)
               << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
+              << convolution_hint
               // Conv2d_2a_3x3
               << ConvolutionLayer(3U, 3U, 32U,
                                   get_weights_accessor(data_path, "/cnn_data/inceptionv4_model/Conv2d_2a_3x3_weights.npy"),
diff --git a/examples/graph_resnet50.cpp b/examples/graph_resnet50.cpp
index 90debb4..a7d7abc 100644
--- a/examples/graph_resnet50.cpp
+++ b/examples/graph_resnet50.cpp
@@ -57,6 +57,8 @@
         bool      enable_tuning            = (target == 2);
         bool      enable_memory_management = true;
 
+        ConvolutionMethod convolution_hint = (target_hint == Target::CL) ? ConvolutionMethod::WINOGRAD : ConvolutionMethod::GEMM;
+
         // Parse arguments
         if(argc < 2)
         {
@@ -97,6 +99,7 @@
                   get_weights_accessor(data_path, "/cnn_data/resnet50_model/conv1_weights.npy"),
                   std::unique_ptr<arm_compute::graph::ITensorAccessor>(nullptr),
                   PadStrideInfo(2, 2, 3, 3))
+              << convolution_hint
               << BatchNormalizationLayer(
                   get_weights_accessor(data_path, "/cnn_data/resnet50_model/conv1_BatchNorm_moving_mean.npy"),
                   get_weights_accessor(data_path, "/cnn_data/resnet50_model/conv1_BatchNorm_moving_variance.npy"),
diff --git a/examples/graph_squeezenet.cpp b/examples/graph_squeezenet.cpp
index b4e00a4..92e6a38 100644
--- a/examples/graph_squeezenet.cpp
+++ b/examples/graph_squeezenet.cpp
@@ -58,6 +58,8 @@
         bool      enable_tuning            = (target == 2);
         bool      enable_memory_management = true;
 
+        ConvolutionMethod convolution_hint = (target_hint == Target::CL) ? ConvolutionMethod::WINOGRAD : ConvolutionMethod::GEMM;
+
         // Parse arguments
         if(argc < 2)
         {
@@ -100,6 +102,7 @@
                   PadStrideInfo(2, 2, 0, 0))
               << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
               << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 3, PadStrideInfo(2, 2, 0, 0, DimensionRoundingType::CEIL)))
+              << convolution_hint
               << ConvolutionLayer(
                   1U, 1U, 16U,
                   get_weights_accessor(data_path, "/cnn_data/squeezenet_v1.0_model/fire2_squeeze1x1_w.npy"),
diff --git a/examples/graph_squeezenet_v1_1.cpp b/examples/graph_squeezenet_v1_1.cpp
index 4ebfd3f..540784e 100644
--- a/examples/graph_squeezenet_v1_1.cpp
+++ b/examples/graph_squeezenet_v1_1.cpp
@@ -61,6 +61,8 @@
         bool      enable_tuning            = (target == 2);
         bool      enable_memory_management = true;
 
+        ConvolutionMethod convolution_hint = (target_hint == Target::CL) ? ConvolutionMethod::WINOGRAD : ConvolutionMethod::GEMM;
+
         // Parse arguments
         if(argc < 2)
         {
@@ -96,6 +98,7 @@
         graph << target_hint
               << InputLayer(TensorDescriptor(TensorShape(227U, 227U, 3U, 1U), DataType::F32),
                             get_input_accessor(image, std::move(preprocessor)))
+              << ConvolutionMethod::DIRECT
               << ConvolutionLayer(
                   3U, 3U, 64U,
                   get_weights_accessor(data_path, "/cnn_data/squeezenet_v1_1_model/conv1_w.npy"),
@@ -103,6 +106,7 @@
                   PadStrideInfo(2, 2, 0, 0))
               << ActivationLayer(ActivationLayerInfo(ActivationLayerInfo::ActivationFunction::RELU))
               << PoolingLayer(PoolingLayerInfo(PoolingType::MAX, 3, PadStrideInfo(2, 2, 0, 0, DimensionRoundingType::CEIL)))
+              << convolution_hint
               << ConvolutionLayer(
                   1U, 1U, 16U,
                   get_weights_accessor(data_path, "/cnn_data/squeezenet_v1_1_model/fire2_squeeze1x1_w.npy"),