COMPMID-452 CL Depthwise Separable Convolution Layer kernel implementation, validation and benchmarking for 3x3xC depthwise filter and DataType::F32.

Change-Id: I95c0c87709763cdbf58d0de66025eac86e30791b
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/82768
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Steven Niu <steven.niu@arm.com>

20 files changed, 1560 insertions, 0 deletions

diff --git a/tests/benchmark_new/CL/DepthwiseConvolution.cpp b/tests/benchmark_new/CL/DepthwiseConvolution.cpp
new file mode 100644
index 0000000000..15c474fd12
--- /dev/null
+++ b/tests/benchmark_new/CL/DepthwiseConvolution.cpp
@@ -0,0 +1,51 @@
+/*
+ * Copyright (c) 2017 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 "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/CLTensorAllocator.h"
+#include "arm_compute/runtime/CL/functions/CLDepthwiseConvolution.h"
+#include "framework/Macros.h"
+#include "framework/datasets/Datasets.h"
+#include "tests/CL/CLAccessor.h"
+#include "tests/TypePrinter.h"
+#include "tests/datasets_new/MobileNetDepthwiseConvolutionDataset.h"
+#include "tests/fixtures_new/DepthwiseConvolutionFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+const auto data_types               = framework::dataset::make("DataType", { DataType::F32 });
+using CLDepthwiseConvolutionFixture = DepthwiseConvolutionFixture<CLTensor, CLDepthwiseConvolution, CLAccessor>;
+
+TEST_SUITE(CL)
+
+REGISTER_FIXTURE_DATA_TEST_CASE(MobileNetDepthwiseConvolution, CLDepthwiseConvolutionFixture, framework::DatasetMode::ALL,
+                                framework::dataset::combine(framework::dataset::combine(datasets::MobileNetDepthwiseConvolutionDataset(), data_types),
+                                                            framework::dataset::make("Batches", { 1, 4, 8 })));
+
+TEST_SUITE_END()
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/benchmark_new/CL/DepthwiseSeparableConvolutionLayer.cpp b/tests/benchmark_new/CL/DepthwiseSeparableConvolutionLayer.cpp
new file mode 100644
index 0000000000..038a2ab72d
--- /dev/null
+++ b/tests/benchmark_new/CL/DepthwiseSeparableConvolutionLayer.cpp
@@ -0,0 +1,51 @@
+/*
+ * Copyright (c) 2017 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 "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/CLTensorAllocator.h"
+#include "arm_compute/runtime/CL/functions/CLDepthwiseSeparableConvolutionLayer.h"
+#include "framework/Macros.h"
+#include "framework/datasets/Datasets.h"
+#include "tests/CL/CLAccessor.h"
+#include "tests/TypePrinter.h"
+#include "tests/datasets_new/MobileNetDepthwiseSeparableConvolutionLayerDataset.h"
+#include "tests/fixtures_new/DepthwiseSeparableConvolutionLayerFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+const auto data_types                             = framework::dataset::make("DataType", { DataType::F32 });
+using CLDepthwiseSeparableConvolutionLayerFixture = DepthwiseSeparableConvolutionLayerFixture<CLTensor, CLDepthwiseSeparableConvolutionLayer, CLAccessor>;
+
+TEST_SUITE(CL)
+
+REGISTER_FIXTURE_DATA_TEST_CASE(MobileNetDepthwiseSeparableConvolutionLayer, CLDepthwiseSeparableConvolutionLayerFixture, framework::DatasetMode::ALL,
+                                framework::dataset::combine(framework::dataset::combine(datasets::MobileNetDepthwiseSeparableConvolutionLayerDataset(), data_types),
+                                                            framework::dataset::make("Batches", { 1, 4, 8 })));
+
+TEST_SUITE_END()
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/datasets_new/DepthwiseConvolutionDataset.h b/tests/datasets_new/DepthwiseConvolutionDataset.h
new file mode 100644
index 0000000000..93da37532f
--- /dev/null
+++ b/tests/datasets_new/DepthwiseConvolutionDataset.h
@@ -0,0 +1,119 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISE_CONVOLUTION_DATASET
+#define ARM_COMPUTE_TEST_DEPTHWISE_CONVOLUTION_DATASET
+
+#include "tests/TypePrinter.h"
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace datasets
+{
+class DepthwiseConvolutionDataset
+{
+public:
+    using type = std::tuple<TensorShape, TensorShape, TensorShape, PadStrideInfo>;
+
+    struct iterator
+    {
+        iterator(std::vector<TensorShape>::const_iterator   src_it,
+                 std::vector<TensorShape>::const_iterator   weights_it,
+                 std::vector<TensorShape>::const_iterator   dst_it,
+                 std::vector<PadStrideInfo>::const_iterator infos_it)
+            : _src_it{ std::move(src_it) },
+              _weights_it{ std::move(weights_it) },
+              _dst_it{ std::move(dst_it) },
+              _infos_it{ std::move(infos_it) }
+        {
+        }
+
+        std::string description() const
+        {
+            std::stringstream description;
+            description << "In=" << *_src_it << ":";
+            description << "Weights=" << *_weights_it << ":";
+            description << "Out=" << *_dst_it << ":";
+            description << "Info=" << *_infos_it;
+            return description.str();
+        }
+
+        DepthwiseConvolutionDataset::type operator*() const
+        {
+            return std::make_tuple(*_src_it, *_weights_it, *_dst_it, *_infos_it);
+        }
+
+        iterator &operator++()
+        {
+            ++_src_it;
+            ++_weights_it;
+            ++_dst_it;
+            ++_infos_it;
+
+            return *this;
+        }
+
+    private:
+        std::vector<TensorShape>::const_iterator   _src_it;
+        std::vector<TensorShape>::const_iterator   _weights_it;
+        std::vector<TensorShape>::const_iterator   _dst_it;
+        std::vector<PadStrideInfo>::const_iterator _infos_it;
+    };
+
+    iterator begin() const
+    {
+        return iterator(_src_shapes.begin(), _weight_shapes.begin(), _dst_shapes.begin(), _infos.begin());
+    }
+
+    int size() const
+    {
+        return std::min(_src_shapes.size(), std::min(_weight_shapes.size(), std::min(_dst_shapes.size(), _infos.size())));
+    }
+
+    void add_config(TensorShape src, TensorShape weights, TensorShape dst, PadStrideInfo info)
+    {
+        _src_shapes.emplace_back(std::move(src));
+        _weight_shapes.emplace_back(std::move(weights));
+        _dst_shapes.emplace_back(std::move(dst));
+        _infos.emplace_back(std::move(info));
+    }
+
+protected:
+    DepthwiseConvolutionDataset()                               = default;
+    DepthwiseConvolutionDataset(DepthwiseConvolutionDataset &&) = default;
+
+private:
+    std::vector<TensorShape>   _src_shapes{};
+    std::vector<TensorShape>   _weight_shapes{};
+    std::vector<TensorShape>   _dst_shapes{};
+    std::vector<PadStrideInfo> _infos{};
+};
+} // namespace datasets
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_DEPTHWISE_CONVOLUTION_DATASET */
diff --git a/tests/datasets_new/DepthwiseSeparableConvolutionLayerDataset.h b/tests/datasets_new/DepthwiseSeparableConvolutionLayerDataset.h
new file mode 100644
index 0000000000..b60002c844
--- /dev/null
+++ b/tests/datasets_new/DepthwiseSeparableConvolutionLayerDataset.h
@@ -0,0 +1,149 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_DATASET
+#define ARM_COMPUTE_TEST_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_DATASET
+
+#include "tests/TypePrinter.h"
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace datasets
+{
+class DepthwiseSeparableConvolutionLayerDataset
+{
+public:
+    using type = std::tuple<TensorShape, TensorShape, TensorShape, TensorShape, TensorShape, TensorShape, PadStrideInfo, PadStrideInfo>;
+
+    struct iterator
+    {
+        iterator(std::vector<TensorShape>::const_iterator   src_it,
+                 std::vector<TensorShape>::const_iterator   filter_it,
+                 std::vector<TensorShape>::const_iterator   depthwise_out_it,
+                 std::vector<TensorShape>::const_iterator   weights_it,
+                 std::vector<TensorShape>::const_iterator   biases_it,
+                 std::vector<TensorShape>::const_iterator   dst_it,
+                 std::vector<PadStrideInfo>::const_iterator depthwise_infos_it,
+                 std::vector<PadStrideInfo>::const_iterator pointwise_infos_it)
+            : _src_it{ std::move(src_it) },
+              _filter_it{ std::move(filter_it) },
+              _depthwise_out_it{ std::move(depthwise_out_it) },
+              _weights_it{ std::move(weights_it) },
+              _biases_it{ std::move(biases_it) },
+              _dst_it{ std::move(dst_it) },
+              _depthwise_infos_it{ std::move(depthwise_infos_it) },
+              _pointwise_infos_it{ std::move(pointwise_infos_it) }
+        {
+        }
+
+        std::string description() const
+        {
+            std::stringstream description;
+            description << "In=" << *_src_it << ":";
+            description << "Filter=" << *_filter_it << ":";
+            description << "Depthwise Out=" << *_depthwise_out_it << ":";
+            description << "Weights=" << *_weights_it << ":";
+            description << "Biases=" << *_biases_it << ":";
+            description << "Out=" << *_dst_it << ":";
+            description << "Depthwise Info=" << *_depthwise_infos_it;
+            description << "Pointwise Info=" << *_pointwise_infos_it;
+            return description.str();
+        }
+
+        DepthwiseSeparableConvolutionLayerDataset::type operator*() const
+        {
+            return std::make_tuple(*_src_it, *_filter_it, *_depthwise_out_it, *_weights_it, *_biases_it, *_dst_it, *_depthwise_infos_it, *_pointwise_infos_it);
+        }
+
+        iterator &operator++()
+        {
+            ++_src_it;
+            ++_filter_it;
+            ++_depthwise_out_it;
+            ++_weights_it;
+            ++_biases_it;
+            ++_dst_it;
+            ++_depthwise_infos_it;
+            ++_pointwise_infos_it;
+
+            return *this;
+        }
+
+    private:
+        std::vector<TensorShape>::const_iterator   _src_it;
+        std::vector<TensorShape>::const_iterator   _filter_it;
+        std::vector<TensorShape>::const_iterator   _depthwise_out_it;
+        std::vector<TensorShape>::const_iterator   _weights_it;
+        std::vector<TensorShape>::const_iterator   _biases_it;
+        std::vector<TensorShape>::const_iterator   _dst_it;
+        std::vector<PadStrideInfo>::const_iterator _depthwise_infos_it;
+        std::vector<PadStrideInfo>::const_iterator _pointwise_infos_it;
+    };
+
+    iterator begin() const
+    {
+        return iterator(_src_shapes.begin(), _filter_shapes.begin(), _depthwise_out_shapes.begin(), _weight_shapes.begin(), _bias_shapes.begin(), _dst_shapes.begin(), _depthwise_infos.begin(),
+                        _pointwise_infos.begin());
+    }
+
+    int size() const
+    {
+        return std::min(_src_shapes.size(), std::min(_filter_shapes.size(), std::min(_depthwise_out_shapes.size(), std::min(_weight_shapes.size(), std::min(_bias_shapes.size(), std::min(_dst_shapes.size(),
+                                                                                                                            std::min(_depthwise_infos.size(), _pointwise_infos.size())))))));
+    }
+
+    void add_config(TensorShape src, TensorShape filter, TensorShape depthwise_out, TensorShape weights, TensorShape biases, TensorShape dst, PadStrideInfo depthwise_info, PadStrideInfo pointwise_info)
+    {
+        _src_shapes.emplace_back(std::move(src));
+        _filter_shapes.emplace_back(std::move(filter));
+        _depthwise_out_shapes.emplace_back(std::move(depthwise_out));
+        _weight_shapes.emplace_back(std::move(weights));
+        _bias_shapes.emplace_back(std::move(biases));
+        _dst_shapes.emplace_back(std::move(dst));
+        _depthwise_infos.emplace_back(std::move(depthwise_info));
+        _pointwise_infos.emplace_back(std::move(pointwise_info));
+    }
+
+protected:
+    DepthwiseSeparableConvolutionLayerDataset()                                             = default;
+    DepthwiseSeparableConvolutionLayerDataset(DepthwiseSeparableConvolutionLayerDataset &&) = default;
+
+private:
+    std::vector<TensorShape>   _src_shapes{};
+    std::vector<TensorShape>   _filter_shapes{};
+    std::vector<TensorShape>   _depthwise_out_shapes{};
+    std::vector<TensorShape>   _weight_shapes{};
+    std::vector<TensorShape>   _bias_shapes{};
+    std::vector<TensorShape>   _dst_shapes{};
+    std::vector<PadStrideInfo> _depthwise_infos{};
+    std::vector<PadStrideInfo> _pointwise_infos{};
+};
+} // namespace datasets
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_DATASET */
diff --git a/tests/datasets_new/LargeDepthwiseConvolutionDataset.h b/tests/datasets_new/LargeDepthwiseConvolutionDataset.h
new file mode 100644
index 0000000000..8fd87f780d
--- /dev/null
+++ b/tests/datasets_new/LargeDepthwiseConvolutionDataset.h
@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) 2017 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_LARGE_DEPTHWISE_CONVOLUTION_DATASET
+#define ARM_COMPUTE_TEST_LARGE_DEPTHWISE_CONVOLUTION_DATASET
+
+#include "tests/datasets_new/DepthwiseConvolutionDataset.h"
+
+#include "tests/TypePrinter.h"
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace datasets
+{
+class LargeDepthwiseConvolutionDataset final : public DepthwiseConvolutionDataset
+{
+public:
+    LargeDepthwiseConvolutionDataset()
+    {
+        add_config(TensorShape(233U, 277U, 55U), TensorShape(3U, 3U, 55U), TensorShape(116U, 275U, 55U), PadStrideInfo(2, 1, 0, 0));
+        add_config(TensorShape(333U, 277U, 77U), TensorShape(3U, 3U, 77U), TensorShape(111U, 138U, 77U), PadStrideInfo(3, 2, 1, 0));
+        add_config(TensorShape(177U, 311U, 22U), TensorShape(3U, 3U, 22U), TensorShape(177U, 156U, 22U), PadStrideInfo(1, 2, 1, 1));
+        add_config(TensorShape(233U, 277U, 55U), TensorShape(3U, 3U, 55U), TensorShape(231U, 138U, 55U), PadStrideInfo(1, 2, 0, 0));
+        add_config(TensorShape(333U, 277U, 77U), TensorShape(3U, 3U, 77U), TensorShape(166U, 93U, 77U), PadStrideInfo(2, 3, 0, 1));
+        add_config(TensorShape(177U, 311U, 22U), TensorShape(3U, 3U, 22U), TensorShape(89U, 311U, 22U), PadStrideInfo(2, 1, 1, 1));
+    }
+};
+} // namespace datasets
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_LARGE_DEPTHWISE_CONVOLUTION_DATASET */
diff --git a/tests/datasets_new/MobileNetDepthwiseConvolutionDataset.h b/tests/datasets_new/MobileNetDepthwiseConvolutionDataset.h
new file mode 100644
index 0000000000..377bee6665
--- /dev/null
+++ b/tests/datasets_new/MobileNetDepthwiseConvolutionDataset.h
@@ -0,0 +1,58 @@
+/*
+ * Copyright (c) 2017 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_MOBILENET_DEPTHWISE_CONVOLUTION_DATASET
+#define ARM_COMPUTE_TEST_MOBILENET_DEPTHWISE_CONVOLUTION_DATASET
+
+#include "tests/datasets_new/DepthwiseConvolutionDataset.h"
+
+#include "tests/TypePrinter.h"
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace datasets
+{
+class MobileNetDepthwiseConvolutionDataset final : public DepthwiseConvolutionDataset
+{
+public:
+    MobileNetDepthwiseConvolutionDataset()
+    {
+        add_config(TensorShape(7U, 7U, 1024U), TensorShape(3U, 3U, 1024U), TensorShape(3U, 3U, 1024U), PadStrideInfo(2, 2, 1, 1));
+        add_config(TensorShape(14U, 14U, 512U), TensorShape(3U, 3U, 512U), TensorShape(7U, 7U, 512U), PadStrideInfo(2, 2, 1, 1));
+        add_config(TensorShape(28U, 28U, 256U), TensorShape(3U, 3U, 256U), TensorShape(14U, 14U, 256U), PadStrideInfo(2, 2, 1, 1));
+        add_config(TensorShape(28U, 28U, 256U), TensorShape(3U, 3U, 256U), TensorShape(28U, 28U, 256U), PadStrideInfo(1, 1, 1, 1));
+        add_config(TensorShape(56U, 56U, 128U), TensorShape(3U, 3U, 128U), TensorShape(28U, 28U, 128U), PadStrideInfo(2, 2, 1, 1));
+        add_config(TensorShape(56U, 56U, 128U), TensorShape(3U, 3U, 128U), TensorShape(56U, 56U, 128U), PadStrideInfo(1, 1, 1, 1));
+        add_config(TensorShape(112U, 112U, 64U), TensorShape(3U, 3U, 64U), TensorShape(56U, 56U, 64U), PadStrideInfo(2, 2, 1, 1));
+        add_config(TensorShape(112U, 112U, 32U), TensorShape(3U, 3U, 32U), TensorShape(112U, 112U, 32U), PadStrideInfo(1, 1, 1, 1));
+    }
+};
+} // namespace datasets
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_MOBILENET_DEPTHWISE_CONVOLUTION_DATASET */
diff --git a/tests/datasets_new/MobileNetDepthwiseSeparableConvolutionLayerDataset.h b/tests/datasets_new/MobileNetDepthwiseSeparableConvolutionLayerDataset.h
new file mode 100644
index 0000000000..d0d3bb3351
--- /dev/null
+++ b/tests/datasets_new/MobileNetDepthwiseSeparableConvolutionLayerDataset.h
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2017 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_MOBILENET_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_DATASET
+#define ARM_COMPUTE_TEST_MOBILENET_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_DATASET
+
+#include "tests/datasets_new/DepthwiseSeparableConvolutionLayerDataset.h"
+
+#include "tests/TypePrinter.h"
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace datasets
+{
+class MobileNetDepthwiseSeparableConvolutionLayerDataset final : public DepthwiseSeparableConvolutionLayerDataset
+{
+public:
+    MobileNetDepthwiseSeparableConvolutionLayerDataset()
+    {
+        add_config(TensorShape(14U, 14U, 512U), TensorShape(3U, 3U, 512U), TensorShape(14U, 14U, 512U), TensorShape(1U, 1U, 512U, 512U), TensorShape(512U), TensorShape(14U, 14U, 512U), PadStrideInfo(1, 1, 1,
+                   1,
+                   DimensionRoundingType::FLOOR),
+                   PadStrideInfo(1, 1, 0, 0, DimensionRoundingType::FLOOR));
+        add_config(TensorShape(14U, 14U, 512U), TensorShape(3U, 3U, 512U), TensorShape(14U, 14U, 512U), TensorShape(1U, 1U, 512U, 512U), TensorShape(512U), TensorShape(14U, 14U, 512U), PadStrideInfo(1, 1, 1,
+                   1,
+                   DimensionRoundingType::FLOOR),
+                   PadStrideInfo(1, 1, 0, 0, DimensionRoundingType::FLOOR));
+        add_config(TensorShape(14U, 14U, 512U), TensorShape(3U, 3U, 512U), TensorShape(14U, 14U, 512U), TensorShape(1U, 1U, 512U, 512U), TensorShape(512U), TensorShape(14U, 14U, 512U), PadStrideInfo(1, 1, 1,
+                   1,
+                   DimensionRoundingType::FLOOR),
+                   PadStrideInfo(1, 1, 0, 0, DimensionRoundingType::FLOOR));
+        add_config(TensorShape(14U, 14U, 512U), TensorShape(3U, 3U, 512U), TensorShape(14U, 14U, 512U), TensorShape(1U, 1U, 512U, 512U), TensorShape(512U), TensorShape(14U, 14U, 512U), PadStrideInfo(1, 1, 1,
+                   1,
+                   DimensionRoundingType::FLOOR),
+                   PadStrideInfo(1, 1, 0, 0, DimensionRoundingType::FLOOR));
+        add_config(TensorShape(14U, 14U, 512U), TensorShape(3U, 3U, 512U), TensorShape(14U, 14U, 512U), TensorShape(1U, 1U, 512U, 512U), TensorShape(512U), TensorShape(14U, 14U, 512U), PadStrideInfo(1, 1, 1,
+                   1,
+                   DimensionRoundingType::FLOOR),
+                   PadStrideInfo(1, 1, 0, 0, DimensionRoundingType::FLOOR));
+    }
+};
+} // namespace datasets
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_MOBILENET_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_DATASET */
diff --git a/tests/datasets_new/SmallDepthwiseConvolutionDataset.h b/tests/datasets_new/SmallDepthwiseConvolutionDataset.h
new file mode 100644
index 0000000000..364673a256
--- /dev/null
+++ b/tests/datasets_new/SmallDepthwiseConvolutionDataset.h
@@ -0,0 +1,56 @@
+/*
+ * Copyright (c) 2017 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_SMALL_DEPTHWISE_CONVOLUTION_DATASET
+#define ARM_COMPUTE_TEST_SMALL_DEPTHWISE_CONVOLUTION_DATASET
+
+#include "tests/datasets_new/DepthwiseConvolutionDataset.h"
+
+#include "tests/TypePrinter.h"
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace datasets
+{
+class SmallDepthwiseConvolutionDataset final : public DepthwiseConvolutionDataset
+{
+public:
+    SmallDepthwiseConvolutionDataset()
+    {
+        add_config(TensorShape(23U, 27U, 5U), TensorShape(3U, 3U, 5U), TensorShape(11U, 25U, 5U), PadStrideInfo(2, 1, 0, 0));
+        add_config(TensorShape(33U, 27U, 7U), TensorShape(3U, 3U, 7U), TensorShape(11U, 13U, 7U), PadStrideInfo(3, 2, 1, 0));
+        add_config(TensorShape(17U, 31U, 2U), TensorShape(3U, 3U, 2U), TensorShape(17U, 16U, 2U), PadStrideInfo(1, 2, 1, 1));
+        add_config(TensorShape(23U, 27U, 5U), TensorShape(3U, 3U, 5U), TensorShape(21U, 13U, 5U), PadStrideInfo(1, 2, 0, 0));
+        add_config(TensorShape(33U, 27U, 7U), TensorShape(3U, 3U, 7U), TensorShape(16U, 9U, 7U), PadStrideInfo(2, 3, 0, 1));
+        add_config(TensorShape(17U, 31U, 2U), TensorShape(3U, 3U, 2U), TensorShape(9U, 31U, 2U), PadStrideInfo(2, 1, 1, 1));
+    }
+};
+} // namespace datasets
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_SMALL_DEPTHWISE_CONVOLUTION_DATASET */
diff --git a/tests/fixtures_new/DepthwiseConvolutionFixture.h b/tests/fixtures_new/DepthwiseConvolutionFixture.h
new file mode 100644
index 0000000000..8223c210d4
--- /dev/null
+++ b/tests/fixtures_new/DepthwiseConvolutionFixture.h
@@ -0,0 +1,88 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISECONVOLUTIONFIXTURE
+#define ARM_COMPUTE_TEST_DEPTHWISECONVOLUTIONFIXTURE
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+#include "framework/Fixture.h"
+#include "tests/Globals.h"
+#include "tests/Utils.h"
+
+namespace arm_compute
+{
+namespace test
+{
+/** Fixture that can be used for NEON and CL */
+template <typename TensorType, typename Function, typename Accessor>
+class DepthwiseConvolutionFixture : public framework::Fixture
+{
+public:
+    template <typename...>
+    void setup(TensorShape src_shape, TensorShape weights_shape, TensorShape dst_shape, PadStrideInfo info, DataType data_type, int batches)
+    {
+        // Set batched in source and destination shapes
+        const unsigned int fixed_point_position = 4;
+        src_shape.set(3 /* batch */, batches);
+        dst_shape.set(3 /* batch */, batches);
+
+        // Create tensors
+        src     = create_tensor<TensorType>(src_shape, data_type, 1, fixed_point_position);
+        weights = create_tensor<TensorType>(weights_shape, data_type, 1, fixed_point_position);
+        dst     = create_tensor<TensorType>(dst_shape, data_type, 1, fixed_point_position);
+
+        // Create and configure function
+        depth_conv.configure(&src, &dst, &weights, info);
+
+        // Allocate tensors
+        src.allocator()->allocate();
+        weights.allocator()->allocate();
+        dst.allocator()->allocate();
+
+        // Fill tensors
+        library->fill_tensor_uniform(Accessor(src), 0);
+        library->fill_tensor_uniform(Accessor(weights), 1);
+    }
+
+    void run()
+    {
+        depth_conv.run();
+    }
+
+    void teardown()
+    {
+        src.allocator()->free();
+        weights.allocator()->free();
+        dst.allocator()->free();
+    }
+
+private:
+    TensorType src{};
+    TensorType weights{};
+    TensorType dst{};
+    Function   depth_conv{};
+};
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_DEPTHWISECONVOLUTIONFIXTURE */
diff --git a/tests/fixtures_new/DepthwiseSeparableConvolutionLayerFixture.h b/tests/fixtures_new/DepthwiseSeparableConvolutionLayerFixture.h
new file mode 100644
index 0000000000..c4ace2ba1c
--- /dev/null
+++ b/tests/fixtures_new/DepthwiseSeparableConvolutionLayerFixture.h
@@ -0,0 +1,103 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISESEPARABLECONVOLUTIONLAYERFIXTURE
+#define ARM_COMPUTE_TEST_DEPTHWISESEPARABLECONVOLUTIONLAYERFIXTURE
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+#include "framework/Fixture.h"
+#include "tests/Globals.h"
+#include "tests/Utils.h"
+
+namespace arm_compute
+{
+namespace test
+{
+/** Fixture that can be used for NEON and CL */
+template <typename TensorType, typename Function, typename Accessor>
+class DepthwiseSeparableConvolutionLayerFixture : public framework::Fixture
+{
+public:
+    template <typename...>
+    void setup(TensorShape src_shape, TensorShape depthwise_weights_shape, TensorShape depthwise_out_shape, TensorShape pointwise_weights_shape, TensorShape biases_shape, TensorShape dst_shape,
+               PadStrideInfo pad_stride_depthwise_info, PadStrideInfo pad_stride_pointwise_info, DataType data_type, int batches)
+    {
+        // Set batched in source and destination shapes
+        const unsigned int fixed_point_position = 4;
+        src_shape.set(3 /* batch */, batches);
+        depthwise_out_shape.set(3 /* batch */, batches);
+        dst_shape.set(3 /* batch */, batches);
+
+        src               = create_tensor<TensorType>(src_shape, data_type, 1, fixed_point_position);
+        depthwise_weights = create_tensor<TensorType>(depthwise_weights_shape, data_type, 1, fixed_point_position);
+        depthwise_out     = create_tensor<TensorType>(depthwise_out_shape, data_type, 1, fixed_point_position);
+        pointwise_weights = create_tensor<TensorType>(pointwise_weights_shape, data_type, 1, fixed_point_position);
+        biases            = create_tensor<TensorType>(biases_shape, data_type, 1, fixed_point_position);
+        dst               = create_tensor<TensorType>(dst_shape, data_type, 1, fixed_point_position);
+
+        // Create and configure function
+        depth_sep_conv_layer.configure(&src, &depthwise_weights, &depthwise_out, &pointwise_weights, &biases, &dst, pad_stride_depthwise_info, pad_stride_pointwise_info);
+
+        // Allocate tensors
+        src.allocator()->allocate();
+        depthwise_weights.allocator()->allocate();
+        depthwise_out.allocator()->allocate();
+        pointwise_weights.allocator()->allocate();
+        biases.allocator()->allocate();
+        dst.allocator()->allocate();
+
+        // Fill tensors
+        library->fill_tensor_uniform(Accessor(src), 0);
+        library->fill_tensor_uniform(Accessor(depthwise_weights), 1);
+        library->fill_tensor_uniform(Accessor(pointwise_weights), 2);
+        library->fill_tensor_uniform(Accessor(biases), 3);
+    }
+
+    void run()
+    {
+        depth_sep_conv_layer.run();
+    }
+
+    void teardown()
+    {
+        src.allocator()->free();
+        depthwise_weights.allocator()->free();
+        depthwise_out.allocator()->free();
+        pointwise_weights.allocator()->free();
+        biases.allocator()->free();
+        dst.allocator()->free();
+    }
+
+private:
+    TensorType src{};
+    TensorType depthwise_weights{};
+    TensorType depthwise_out{};
+    TensorType pointwise_weights{};
+    TensorType biases{};
+    TensorType dst{};
+    Function   depth_sep_conv_layer{};
+};
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_DEPTHWISESEPARABLECONVOLUTIONLAYERFIXTURE */
diff --git a/tests/validation_new/CL/DepthwiseConvolution.cpp b/tests/validation_new/CL/DepthwiseConvolution.cpp
new file mode 100644
index 0000000000..3fda275715
--- /dev/null
+++ b/tests/validation_new/CL/DepthwiseConvolution.cpp
@@ -0,0 +1,69 @@
+/*
+ * Copyright (c) 2017 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 CONCLCTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/CLTensorAllocator.h"
+#include "arm_compute/runtime/CL/functions/CLDepthwiseConvolution.h"
+#include "framework/Asserts.h"
+#include "framework/Macros.h"
+#include "framework/datasets/Datasets.h"
+#include "tests/CL/CLAccessor.h"
+#include "tests/PaddingCalculator.h"
+#include "tests/datasets_new/LargeDepthwiseConvolutionDataset.h"
+#include "tests/datasets_new/SmallDepthwiseConvolutionDataset.h"
+#include "tests/validation_new/Validation.h"
+#include "tests/validation_new/fixtures/DepthwiseConvolutionFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+constexpr AbsoluteTolerance<float> tolerance_f32(0.001f); /**< Tolerance value for comparing reference's output against implementation's output for DataType::F32 */
+} // namespace
+
+TEST_SUITE(CL)
+TEST_SUITE(DepthwiseConvolution)
+
+template <typename T>
+using CLDepthwiseConvolutionFixture = DepthwiseConvolutionValidationFixture<CLTensor, CLAccessor, CLDepthwiseConvolution, T>;
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthwiseConvolutionFixture<float>, framework::DatasetMode::PRECOMMIT, datasets::SmallDepthwiseConvolutionDataset())
+{
+    validate(CLAccessor(_target), _reference, tolerance_f32);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLDepthwiseConvolutionFixture<float>, framework::DatasetMode::NIGHTLY, datasets::LargeDepthwiseConvolutionDataset())
+{
+    validate(CLAccessor(_target), _reference, tolerance_f32);
+}
+
+TEST_SUITE_END()
+TEST_SUITE_END()
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation_new/CL/DepthwiseSeparableConvolutionLayer.cpp b/tests/validation_new/CL/DepthwiseSeparableConvolutionLayer.cpp
new file mode 100644
index 0000000000..e577f5af04
--- /dev/null
+++ b/tests/validation_new/CL/DepthwiseSeparableConvolutionLayer.cpp
@@ -0,0 +1,64 @@
+/*
+ * Copyright (c) 2017 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 CONCLCTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/CLTensor.h"
+#include "arm_compute/runtime/CL/CLTensorAllocator.h"
+#include "arm_compute/runtime/CL/functions/CLDepthwiseSeparableConvolutionLayer.h"
+#include "framework/Asserts.h"
+#include "framework/Macros.h"
+#include "framework/datasets/Datasets.h"
+#include "tests/CL/CLAccessor.h"
+#include "tests/PaddingCalculator.h"
+#include "tests/datasets_new/MobileNetDepthwiseSeparableConvolutionLayerDataset.h"
+#include "tests/validation_new/Validation.h"
+#include "tests/validation_new/fixtures/DepthwiseSeparableConvolutionLayerFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+constexpr AbsoluteTolerance<float> tolerance_f32(0.001f); /**< Tolerance value for comparing reference's output against implementation's output for DataType::F32 */
+} // namespace
+
+TEST_SUITE(CL)
+TEST_SUITE(DepthwiseSeparableConvolutionLayer)
+
+// Configuration test to do
+
+template <typename T>
+using CLDepthwiseSeparableConvolutionLayerFixture = DepthwiseSeparableConvolutionValidationFixture<CLTensor, CLAccessor, CLDepthwiseSeparableConvolutionLayer, T>;
+FIXTURE_DATA_TEST_CASE(RunSmall, CLDepthwiseSeparableConvolutionLayerFixture<float>, framework::DatasetMode::PRECOMMIT, datasets::MobileNetDepthwiseSeparableConvolutionLayerDataset())
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference, tolerance_f32);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation_new/CPP/DepthwiseConvolution.cpp b/tests/validation_new/CPP/DepthwiseConvolution.cpp
new file mode 100644
index 0000000000..be18ffb911
--- /dev/null
+++ b/tests/validation_new/CPP/DepthwiseConvolution.cpp
@@ -0,0 +1,103 @@
+/*
+ * Copyright (c) 2017 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 "DepthwiseConvolution.h"
+
+#include "ConvolutionLayer.h"
+#include "TensorElementAt.h"
+
+#include "tests/validation_new/Helpers.h"
+#include "tests/validation_new/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+/** Perform a depthwise convolution
+ *
+ * - Three dimensions tensors
+ * - Third dimention is number of channels
+ * - Depths of input tensor and filter are equals
+ * - Padding, stride and output shape "match"
+ *
+ */
+template <typename T>
+SimpleTensor<T> depthwise_convolution(const SimpleTensor<T> &src, const SimpleTensor<T> &weights, const TensorShape &dst_shape, const PadStrideInfo &conv_info)
+{
+    // Create reference
+    SimpleTensor<T> dst{ dst_shape, src.data_type(), 1, src.fixed_point_position() };
+
+    // Compute reference
+    const size_t filter_width  = weights.shape().x();
+    const size_t filter_height = weights.shape().y();
+    const size_t filter_plane  = filter_width * filter_height;
+    const size_t input_width   = src.shape().x();
+    const size_t input_height  = src.shape().y();
+    const size_t input_depth   = src.shape().z();
+
+    const size_t filter_half_size = filter_width / 2;
+    const size_t pad_x            = std::min(filter_half_size, static_cast<size_t>(conv_info.pad().first));
+    const size_t pad_y            = std::min(filter_half_size, static_cast<size_t>(conv_info.pad().second));
+    const size_t minimum_x        = -pad_x + filter_half_size;
+    const size_t minimum_y        = -pad_y + filter_half_size;
+
+    int out_pos = 0;
+    for(size_t z = 0; z < input_depth; ++z)
+    {
+        for(size_t y = minimum_y; y < input_height + pad_y - filter_half_size; y += conv_info.stride().second)
+        {
+            for(size_t x = minimum_x; x < input_width + pad_x - filter_half_size; x += conv_info.stride().first)
+            {
+                Coordinates coords(static_cast<int>(x), static_cast<int>(y), static_cast<int>(z));
+                size_t      filter_offset = filter_plane * z;
+
+                T val = 0;
+                for(int j = y - filter_half_size; j <= static_cast<int>(y + filter_half_size); ++j)
+                {
+                    for(int i = x - filter_half_size; i <= static_cast<int>(x + filter_half_size); ++i)
+                    {
+                        coords.set(0, i);
+                        coords.set(1, j);
+                        val += *(weights.data() + filter_offset) * tensor_elem_at(src, coords, BorderMode::CONSTANT, 0.f);
+                        ++filter_offset;
+                    }
+                }
+                coords.set(0, x);
+                coords.set(1, y);
+                dst[out_pos++] = saturate_cast<T>(val);
+            }
+        }
+    }
+
+    return dst;
+}
+
+template SimpleTensor<float> depthwise_convolution(const SimpleTensor<float> &src, const SimpleTensor<float> &weights, const TensorShape &dst_shape, const PadStrideInfo &conv_info);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation_new/CPP/DepthwiseConvolution.h b/tests/validation_new/CPP/DepthwiseConvolution.h
new file mode 100644
index 0000000000..4a92620880
--- /dev/null
+++ b/tests/validation_new/CPP/DepthwiseConvolution.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISE_CONVOLUTION_H__
+#define __ARM_COMPUTE_TEST_DEPTHWISE_CONVOLUTION_H__
+
+#include "tests/SimpleTensor.h"
+#include "tests/validation_new/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template <typename T>
+SimpleTensor<T> depthwise_convolution(const SimpleTensor<T> &src, const SimpleTensor<T> &weights, const TensorShape &dst_shape, const PadStrideInfo &conv_info);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_TEST_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_H__ */
diff --git a/tests/validation_new/CPP/DepthwiseSeparableConvolutionLayer.cpp b/tests/validation_new/CPP/DepthwiseSeparableConvolutionLayer.cpp
new file mode 100644
index 0000000000..88584275cf
--- /dev/null
+++ b/tests/validation_new/CPP/DepthwiseSeparableConvolutionLayer.cpp
@@ -0,0 +1,60 @@
+/*
+ * Copyright (c) 2017 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 "DepthwiseConvolution.h"
+
+#include "DepthwiseSeparableConvolutionLayer.h"
+
+#include "ConvolutionLayer.h"
+#include "TensorElementAt.h"
+
+#include "tests/validation_new/Helpers.h"
+#include "tests/validation_new/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+// Depthwise separable convolution layer
+template <typename T>
+SimpleTensor<T> depthwise_separable_convolution_layer(const SimpleTensor<T> &src, const SimpleTensor<T> &depthwise_weights, const TensorShape &depthwise_out_shape,
+                                                      const SimpleTensor<T> &pointwise_weights,
+                                                      const SimpleTensor<T> &biases, const TensorShape &dst_shape, const PadStrideInfo &depthwise_conv_info, const PadStrideInfo &pointwise_conv_info)
+{
+    // Compute reference
+    SimpleTensor<T> depthwise_out = depthwise_convolution(src, depthwise_weights, depthwise_out_shape, depthwise_conv_info);
+    SimpleTensor<T> dst           = convolution_layer(depthwise_out, pointwise_weights, biases, dst_shape, pointwise_conv_info);
+
+    return dst;
+}
+
+template SimpleTensor<float> depthwise_separable_convolution_layer(const SimpleTensor<float> &in, const SimpleTensor<float> &depthwise_weights, const TensorShape &depthwise_out_shape,
+                                                                   const SimpleTensor<float> &pointwise_weights, const SimpleTensor<float> &biases, const TensorShape &dst_shape, const PadStrideInfo &depthwise_conv_info, const PadStrideInfo &pointwise_conv_info);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation_new/CPP/DepthwiseSeparableConvolutionLayer.h b/tests/validation_new/CPP/DepthwiseSeparableConvolutionLayer.h
new file mode 100644
index 0000000000..0c9f118b84
--- /dev/null
+++ b/tests/validation_new/CPP/DepthwiseSeparableConvolutionLayer.h
@@ -0,0 +1,46 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_H__
+#define __ARM_COMPUTE_TEST_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_H__
+
+#include "tests/SimpleTensor.h"
+#include "tests/validation_new/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template <typename T>
+SimpleTensor<T> depthwise_separable_convolution_layer(const SimpleTensor<T> &src, const SimpleTensor<T> &depthwise_weights, const TensorShape &depthwise_out_shape,
+                                                      const SimpleTensor<T> &pointwise_weights,
+                                                      const SimpleTensor<T> &biases, const TensorShape &dst_shape, const PadStrideInfo &depthwise_conv_info, const PadStrideInfo &pointwise_conv_info);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_TEST_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_H__ */
diff --git a/tests/validation_new/CPP/TensorElementAt.cpp b/tests/validation_new/CPP/TensorElementAt.cpp
new file mode 100644
index 0000000000..ca4cca9650
--- /dev/null
+++ b/tests/validation_new/CPP/TensorElementAt.cpp
@@ -0,0 +1,70 @@
+/*
+ * Copyright (c) 2017 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 "TensorElementAt.h"
+
+#include "tests/validation_new/Helpers.h"
+#include "tests/validation_new/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+// Return a tensor element at a specified coordinate with different border modes
+template <typename T>
+T tensor_elem_at(const SimpleTensor<T> &in, Coordinates coord, BorderMode border_mode, T constant_border_value)
+{
+    const int  x      = coord.x();
+    const int  y      = coord.y();
+    const auto width  = static_cast<int>(in.shape().x());
+    const auto height = static_cast<int>(in.shape().y());
+
+    // If coordinates beyond range of tensor's width or height
+    if(x < 0 || y < 0 || x >= width || y >= height)
+    {
+        if(border_mode == BorderMode::REPLICATE)
+        {
+            coord.set(0, std::max(0, std::min(x, width - 1)));
+            coord.set(1, std::max(0, std::min(y, height - 1)));
+
+            return in[coord2index(in.shape(), coord)];
+        }
+        else
+        {
+            return constant_border_value;
+        }
+    }
+    else
+    {
+        return in[coord2index(in.shape(), coord)];
+    }
+}
+template float tensor_elem_at(const SimpleTensor<float> &in, Coordinates coord, BorderMode border_mode, float constant_border_value);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation_new/CPP/TensorElementAt.h b/tests/validation_new/CPP/TensorElementAt.h
new file mode 100644
index 0000000000..d0d68ab6af
--- /dev/null
+++ b/tests/validation_new/CPP/TensorElementAt.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (c) 2017 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_TENSOR_ELEMENT_AT_H__
+#define __ARM_COMPUTE_TEST_TENSOR_ELEMENT_AT_H__
+
+#include "tests/SimpleTensor.h"
+#include "tests/validation_new/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template <typename T>
+T tensor_elem_at(const SimpleTensor<T> &in, Coordinates coord, BorderMode border_mode, T constant_border_value);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_TEST_TENSOR_ELEMENT_AT_H__ */
diff --git a/tests/validation_new/fixtures/DepthwiseConvolutionFixture.h b/tests/validation_new/fixtures/DepthwiseConvolutionFixture.h
new file mode 100644
index 0000000000..b745c7e53c
--- /dev/null
+++ b/tests/validation_new/fixtures/DepthwiseConvolutionFixture.h
@@ -0,0 +1,120 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISE_CONVOLUTION_FIXTURE
+#define ARM_COMPUTE_TEST_DEPTHWISE_CONVOLUTION_FIXTURE
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+#include "framework/Asserts.h"
+#include "framework/Fixture.h"
+#include "tests/AssetsLibrary.h"
+#include "tests/Globals.h"
+#include "tests/IAccessor.h"
+#include "tests/validation_new/CPP/DepthwiseConvolution.h"
+#include "tests/validation_new/Helpers.h"
+
+#include <random>
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class DepthwiseConvolutionValidationFixture : public framework::Fixture
+{
+public:
+    template <typename...>
+    void setup(TensorShape in_shape, TensorShape weights_shape, TensorShape out_shape, PadStrideInfo pad_stride_info)
+    {
+        _target    = compute_target(in_shape, weights_shape, out_shape, pad_stride_info);
+        _reference = compute_reference(in_shape, weights_shape, out_shape, pad_stride_info);
+    }
+
+protected:
+    template <typename U>
+    void fill(U &&tensor, int i)
+    {
+        switch(tensor.data_type())
+        {
+            case DataType::F32:
+            {
+                std::uniform_real_distribution<> distribution(-1.0f, 1.0f);
+                library->fill(tensor, distribution, i);
+                break;
+            }
+            default:
+                library->fill_tensor_uniform(tensor, i);
+        }
+    }
+
+    TensorType compute_target(const TensorShape &input_shape, const TensorShape &weights_shape, const TensorShape &output_shape, PadStrideInfo &pad_stride_info)
+    {
+        // Create tensors
+        TensorType src     = create_tensor<TensorType>(input_shape, DataType::F32);
+        TensorType weights = create_tensor<TensorType>(weights_shape, DataType::F32);
+        TensorType dst     = create_tensor<TensorType>(output_shape, DataType::F32);
+
+        // Create Depthwise Convolution configure function
+        CLDepthwiseConvolution depthwise_convolution;
+        depthwise_convolution.configure(&src, &dst, &weights, pad_stride_info);
+
+        // Allocate tensors
+        src.allocator()->allocate();
+        weights.allocator()->allocate();
+        dst.allocator()->allocate();
+
+        ARM_COMPUTE_EXPECT(!src.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!weights.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!dst.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Fill tensors
+        fill(AccessorType(src), 0);
+        fill(AccessorType(weights), 1);
+
+        // Compute function
+        depthwise_convolution.run();
+
+        return dst;
+    }
+
+    SimpleTensor<T> compute_reference(const TensorShape &in_shape, const TensorShape &weights_shape, const TensorShape &out_shape, const PadStrideInfo &pad_stride_info)
+    {
+        SimpleTensor<T> src(in_shape, DataType::F32);
+        SimpleTensor<T> weights(weights_shape, DataType::F32);
+
+        fill(src, 0);
+        fill(weights, 1);
+
+        return reference::depthwise_convolution(src, weights, out_shape, pad_stride_info);
+    }
+
+    TensorType      _target{};
+    SimpleTensor<T> _reference{};
+};
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_DEPTHWISE_CONVOLUTION_FIXTURE */
diff --git a/tests/validation_new/fixtures/DepthwiseSeparableConvolutionLayerFixture.h b/tests/validation_new/fixtures/DepthwiseSeparableConvolutionLayerFixture.h
new file mode 100644
index 0000000000..c2578a9ce0
--- /dev/null
+++ b/tests/validation_new/fixtures/DepthwiseSeparableConvolutionLayerFixture.h
@@ -0,0 +1,139 @@
+/*
+ * Copyright (c) 2017 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_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_FIXTURE
+#define ARM_COMPUTE_TEST_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_FIXTURE
+
+#include "arm_compute/core/TensorShape.h"
+#include "arm_compute/core/Types.h"
+#include "framework/Asserts.h"
+#include "framework/Fixture.h"
+#include "tests/AssetsLibrary.h"
+#include "tests/Globals.h"
+#include "tests/IAccessor.h"
+#include "tests/validation_new/CPP/DepthwiseSeparableConvolutionLayer.h"
+#include "tests/validation_new/Helpers.h"
+
+#include <random>
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class DepthwiseSeparableConvolutionValidationFixture : public framework::Fixture
+{
+public:
+    template <typename...>
+    void setup(TensorShape in_shape, TensorShape depthwise_weights_shape, TensorShape depthwise_out_shape, TensorShape pointwise_weights_shape, TensorShape biases_shape, TensorShape output_shape,
+               PadStrideInfo pad_stride_depthwise_info, PadStrideInfo pad_stride_pointwise_info)
+    {
+        _target    = compute_target(in_shape, depthwise_weights_shape, depthwise_out_shape, pointwise_weights_shape, biases_shape, output_shape, pad_stride_depthwise_info, pad_stride_pointwise_info);
+        _reference = compute_reference(in_shape, depthwise_weights_shape, depthwise_out_shape, pointwise_weights_shape, biases_shape, output_shape, pad_stride_depthwise_info, pad_stride_pointwise_info);
+    }
+
+protected:
+    template <typename U>
+    void fill(U &&tensor, int i)
+    {
+        switch(tensor.data_type())
+        {
+            case DataType::F32:
+            {
+                std::uniform_real_distribution<> distribution(-1.0f, 1.0f);
+                library->fill(tensor, distribution, i);
+                break;
+            }
+            default:
+                library->fill_tensor_uniform(tensor, i);
+        }
+    }
+
+    TensorType compute_target(const TensorShape &input_shape, const TensorShape &depthwise_weights_shape, const TensorShape &depthwise_out_shape, const TensorShape &pointwise_weights_shape,
+                              const TensorShape &biases_shape,
+                              const TensorShape &output_shape, const PadStrideInfo &pad_stride_depthwise_info, const PadStrideInfo &pad_stride_pointwise_info)
+    {
+        // Create tensors
+        TensorType src               = create_tensor<TensorType>(input_shape, DataType::F32);
+        TensorType depthwise_weights = create_tensor<TensorType>(depthwise_weights_shape, DataType::F32);
+        TensorType depthwise_out     = create_tensor<TensorType>(depthwise_out_shape, DataType::F32);
+        TensorType pointwise_weights = create_tensor<TensorType>(pointwise_weights_shape, DataType::F32);
+        TensorType biases            = create_tensor<TensorType>(biases_shape, DataType::F32);
+        TensorType dst               = create_tensor<TensorType>(output_shape, DataType::F32);
+
+        // Create Depthwise Separable Convolution Layer configure function
+        CLDepthwiseSeparableConvolutionLayer depthwise_separable_convolution_layer;
+        depthwise_separable_convolution_layer.configure(&src, &depthwise_weights, &depthwise_out, &pointwise_weights, &biases, &dst, pad_stride_depthwise_info, pad_stride_pointwise_info);
+
+        // Allocate tensors
+        src.allocator()->allocate();
+        depthwise_weights.allocator()->allocate();
+        depthwise_out.allocator()->allocate();
+        pointwise_weights.allocator()->allocate();
+        biases.allocator()->allocate();
+        dst.allocator()->allocate();
+
+        ARM_COMPUTE_EXPECT(!src.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!depthwise_weights.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!depthwise_out.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!pointwise_weights.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!biases.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!dst.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Fill tensors
+        fill(AccessorType(src), 0);
+        fill(AccessorType(depthwise_weights), 1);
+        fill(AccessorType(pointwise_weights), 2);
+        fill(AccessorType(biases), 3);
+
+        // Compute function
+        depthwise_separable_convolution_layer.run();
+
+        return dst;
+    }
+
+    SimpleTensor<T> compute_reference(const TensorShape &in_shape, const TensorShape &depthwise_weights_shape, const TensorShape &depthwise_out_shape, const TensorShape &pointwise_weights_shape,
+                                      const TensorShape &biases_shape, const TensorShape &dst_shape, const PadStrideInfo &pad_stride_depthwise_info, const PadStrideInfo &pad_stride_pointwise_info)
+    {
+        SimpleTensor<T> src(in_shape, DataType::F32);
+        SimpleTensor<T> depthwise_weights(depthwise_weights_shape, DataType::F32);
+        SimpleTensor<T> pointwise_weights(pointwise_weights_shape, DataType::F32);
+        SimpleTensor<T> biases(biases_shape, DataType::F32);
+
+        fill(src, 0);
+        fill(depthwise_weights, 1);
+        fill(pointwise_weights, 2);
+        fill(biases, 3);
+
+        return reference::depthwise_separable_convolution_layer(src, depthwise_weights, depthwise_out_shape, pointwise_weights, biases, dst_shape, pad_stride_depthwise_info, pad_stride_pointwise_info);
+    }
+
+    TensorType      _target{};
+    SimpleTensor<T> _reference{};
+};
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_DEPTHWISE_SEPARABLE_CONVOLUTION_LAYER_FIXTURE */