COMPMID-493: Port ArithmeticSubtraction to new validation

Change-Id: Ia49f337a8a574392ac0948fe65e9dca7a5fac2d0
Reviewed-on: http://mpd-gerrit.cambridge.arm.com/86338
Tested-by: Kaizen <jeremy.johnson+kaizengerrit@arm.com>
Reviewed-by: Anthony Barbier <anthony.barbier@arm.com>

14 files changed, 713 insertions, 673 deletions

diff --git a/tests/validation/CL/ArithmeticSubtraction.cpp b/tests/validation/CL/ArithmeticSubtraction.cpp
new file mode 100644
index 0000000000..e2bc4b2d79
--- /dev/null
+++ b/tests/validation/CL/ArithmeticSubtraction.cpp
@@ -0,0 +1,237 @@
+/*
+ * 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/CLArithmeticSubtraction.h"
+#include "tests/CL/CLAccessor.h"
+#include "tests/PaddingCalculator.h"
+#include "tests/datasets/ConvertPolicyDataset.h"
+#include "tests/datasets/ShapeDatasets.h"
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Macros.h"
+#include "tests/framework/datasets/Datasets.h"
+#include "tests/validation/Validation.h"
+#include "tests/validation/fixtures/ArithmeticSubtractionFixture.h"
+#include "tests/validation/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+/** Input data sets **/
+const auto ArithmeticSubtractionU8Dataset = combine(combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::U8)),
+                                                    framework::dataset::make("DataType",
+                                                                             DataType::U8));
+const auto ArithmeticSubtractionS16Dataset = combine(combine(framework::dataset::make("DataType", { DataType::U8, DataType::S16 }), framework::dataset::make("DataType", DataType::S16)),
+                                                     framework::dataset::make("DataType", DataType::S16));
+const auto ArithmeticSubtractionQS8Dataset = combine(combine(framework::dataset::make("DataType", DataType::QS8), framework::dataset::make("DataType", DataType::QS8)),
+                                                     framework::dataset::make("DataType", DataType::QS8));
+const auto ArithmeticSubtractionQS16Dataset = combine(combine(framework::dataset::make("DataType", DataType::QS16), framework::dataset::make("DataType", DataType::QS16)),
+                                                      framework::dataset::make("DataType", DataType::QS16));
+const auto ArithmeticSubtractionFP16Dataset = combine(combine(framework::dataset::make("DataType", DataType::F16), framework::dataset::make("DataType", DataType::F16)),
+                                                      framework::dataset::make("DataType", DataType::F16));
+const auto ArithmeticSubtractionFP32Dataset = combine(combine(framework::dataset::make("DataType", DataType::F32), framework::dataset::make("DataType", DataType::F32)),
+                                                      framework::dataset::make("DataType", DataType::F32));
+} // namespace
+
+TEST_SUITE(CL)
+TEST_SUITE(ArithmeticSubtraction)
+
+template <typename T>
+using CLArithmeticSubtractionFixture = ArithmeticSubtractionValidationFixture<CLTensor, CLAccessor, CLArithmeticSubtraction, T>;
+
+TEST_SUITE(U8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+               shape, policy)
+{
+    // Create tensors
+    CLTensor ref_src1 = create_tensor<CLTensor>(shape, DataType::U8);
+    CLTensor ref_src2 = create_tensor<CLTensor>(shape, DataType::U8);
+    CLTensor dst      = create_tensor<CLTensor>(shape, DataType::U8);
+
+    // Create and Configure function
+    CLArithmeticSubtraction sub;
+    sub.configure(&ref_src1, &ref_src2, &dst, policy);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(ref_src1.info()->padding(), padding);
+    validate(ref_src2.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLArithmeticSubtractionFixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionU8Dataset),
+                                                                                                                     framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(S16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::U8, DataType::S16 })),
+                                                                   framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+               shape, data_type, policy)
+{
+    // Create tensors
+    CLTensor ref_src1 = create_tensor<CLTensor>(shape, data_type);
+    CLTensor ref_src2 = create_tensor<CLTensor>(shape, DataType::S16);
+    CLTensor dst      = create_tensor<CLTensor>(shape, DataType::S16);
+
+    // Create and Configure function
+    CLArithmeticSubtraction sub;
+    sub.configure(&ref_src1, &ref_src2, &dst, policy);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(ref_src1.info()->padding(), padding);
+    validate(ref_src2.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLArithmeticSubtractionFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionS16Dataset),
+                                                                                                                     framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLArithmeticSubtractionFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(datasets::LargeShapes(), ArithmeticSubtractionS16Dataset),
+                                                                                                                   framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+template <typename T>
+using CLArithmeticSubtractionFixedPointFixture = ArithmeticSubtractionValidationFixedPointFixture<CLTensor, CLAccessor, CLArithmeticSubtraction, T>;
+
+TEST_SUITE(Quantized)
+TEST_SUITE(QS8)
+FIXTURE_DATA_TEST_CASE(RunSmall, CLArithmeticSubtractionFixedPointFixture<int8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), ArithmeticSubtractionQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 7)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLArithmeticSubtractionFixedPointFixture<int8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), ArithmeticSubtractionQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 7)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(QS16)
+FIXTURE_DATA_TEST_CASE(RunSmall, CLArithmeticSubtractionFixedPointFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(),
+                       ArithmeticSubtractionQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 15)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLArithmeticSubtractionFixedPointFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), ArithmeticSubtractionQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 15)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE(Float)
+TEST_SUITE(FP16)
+FIXTURE_DATA_TEST_CASE(RunSmall, CLArithmeticSubtractionFixture<half_float::half>, framework::DatasetMode::ALL, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionFP16Dataset),
+                                                                                                                        framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(FP32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+               shape, policy)
+{
+    // Create tensors
+    CLTensor ref_src1 = create_tensor<CLTensor>(shape, DataType::F32);
+    CLTensor ref_src2 = create_tensor<CLTensor>(shape, DataType::F32);
+    CLTensor dst      = create_tensor<CLTensor>(shape, DataType::F32);
+
+    // Create and Configure function
+    CLArithmeticSubtraction sub;
+    sub.configure(&ref_src1, &ref_src2, &dst, policy);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(ref_src1.info()->padding(), padding);
+    validate(ref_src2.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, CLArithmeticSubtractionFixture<float>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionFP32Dataset),
+                                                                                                                   framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, CLArithmeticSubtractionFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(datasets::LargeShapes(), ArithmeticSubtractionFP32Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE_END()
+TEST_SUITE_END()
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/CPP/ArithmeticSubtraction.cpp b/tests/validation/CPP/ArithmeticSubtraction.cpp
new file mode 100644
index 0000000000..fa7fec9d1b
--- /dev/null
+++ b/tests/validation/CPP/ArithmeticSubtraction.cpp
@@ -0,0 +1,63 @@
+/*
+ * 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 "ArithmeticSubtraction.h"
+
+#include "tests/validation/FixedPoint.h"
+#include "tests/validation/Helpers.h"
+#include "tests/validation/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template <typename T>
+SimpleTensor<T> arithmetic_subtraction(const SimpleTensor<T> &src1, const SimpleTensor<T> &src2, DataType dst_data_type, ConvertPolicy convert_policy)
+{
+    SimpleTensor<T> result(src1.shape(), dst_data_type);
+
+    using intermediate_type = typename common_promoted_signed_type<T>::intermediate_type;
+
+    for(int i = 0; i < src1.num_elements(); ++i)
+    {
+        intermediate_type val = static_cast<intermediate_type>(src1[i]) - static_cast<intermediate_type>(src2[i]);
+        result[i]             = (convert_policy == ConvertPolicy::SATURATE) ? saturate_cast<T>(val) : static_cast<T>(val);
+    }
+
+    return result;
+}
+
+template SimpleTensor<uint8_t> arithmetic_subtraction(const SimpleTensor<uint8_t> &src1, const SimpleTensor<uint8_t> &src2, DataType dst_data_type, ConvertPolicy convert_policy);
+template SimpleTensor<int16_t> arithmetic_subtraction(const SimpleTensor<int16_t> &src1, const SimpleTensor<int16_t> &src2, DataType dst_data_type, ConvertPolicy convert_policy);
+template SimpleTensor<int8_t> arithmetic_subtraction(const SimpleTensor<int8_t> &src1, const SimpleTensor<int8_t> &src2, DataType dst_data_type, ConvertPolicy convert_policy);
+template SimpleTensor<half_float::half> arithmetic_subtraction(const SimpleTensor<half_float::half> &src1, const SimpleTensor<half_float::half> &src2, DataType dst_data_type,
+                                                               ConvertPolicy convert_policy);
+template SimpleTensor<float> arithmetic_subtraction(const SimpleTensor<float> &src1, const SimpleTensor<float> &src2, DataType dst_data_type, ConvertPolicy convert_policy);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/CPP/ArithmeticSubtraction.h b/tests/validation/CPP/ArithmeticSubtraction.h
new file mode 100644
index 0000000000..18b0d121a0
--- /dev/null
+++ b/tests/validation/CPP/ArithmeticSubtraction.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_ARITHMETIC_SUBTRACTION_H__
+#define __ARM_COMPUTE_TEST_ARITHMETIC_SUBTRACTION_H__
+
+#include "tests/SimpleTensor.h"
+#include "tests/validation/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace reference
+{
+template <typename T>
+SimpleTensor<T> arithmetic_subtraction(const SimpleTensor<T> &src1, const SimpleTensor<T> &src2, DataType dst_data_type, ConvertPolicy convert_policy);
+} // namespace reference
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* __ARM_COMPUTE_TEST_ARITHMETIC_SUBTRACTION_H__ */
diff --git a/tests/validation/NEON/ArithmeticSubtraction.cpp b/tests/validation/NEON/ArithmeticSubtraction.cpp
new file mode 100644
index 0000000000..124f05a470
--- /dev/null
+++ b/tests/validation/NEON/ArithmeticSubtraction.cpp
@@ -0,0 +1,241 @@
+/*
+ * 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/NEON/functions/NEArithmeticSubtraction.h"
+#include "arm_compute/runtime/Tensor.h"
+#include "arm_compute/runtime/TensorAllocator.h"
+#include "tests/NEON/Accessor.h"
+#include "tests/PaddingCalculator.h"
+#include "tests/datasets/ConvertPolicyDataset.h"
+#include "tests/datasets/ShapeDatasets.h"
+#include "tests/framework/Asserts.h"
+#include "tests/framework/Macros.h"
+#include "tests/framework/datasets/Datasets.h"
+#include "tests/validation/Validation.h"
+#include "tests/validation/fixtures/ArithmeticSubtractionFixture.h"
+#include "tests/validation/half.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+/** Input data sets **/
+const auto ArithmeticSubtractionU8Dataset = combine(combine(framework::dataset::make("DataType", DataType::U8), framework::dataset::make("DataType", DataType::U8)),
+                                                    framework::dataset::make("DataType",
+                                                                             DataType::U8));
+const auto ArithmeticSubtractionS16Dataset = combine(combine(framework::dataset::make("DataType", { DataType::U8, DataType::S16 }), framework::dataset::make("DataType", DataType::S16)),
+                                                     framework::dataset::make("DataType", DataType::S16));
+const auto ArithmeticSubtractionQS8Dataset = combine(combine(framework::dataset::make("DataType", DataType::QS8), framework::dataset::make("DataType", DataType::QS8)),
+                                                     framework::dataset::make("DataType", DataType::QS8));
+const auto ArithmeticSubtractionQS16Dataset = combine(combine(framework::dataset::make("DataType", DataType::QS16), framework::dataset::make("DataType", DataType::QS16)),
+                                                      framework::dataset::make("DataType", DataType::QS16));
+#ifdef ARM_COMPUTE_ENABLE_FP16
+const auto ArithmeticSubtractionFP16Dataset = combine(combine(framework::dataset::make("DataType", DataType::F16), framework::dataset::make("DataType", DataType::F16)),
+                                                      framework::dataset::make("DataType", DataType::F16));
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
+const auto ArithmeticSubtractionFP32Dataset = combine(combine(framework::dataset::make("DataType", DataType::F32), framework::dataset::make("DataType", DataType::F32)),
+                                                      framework::dataset::make("DataType", DataType::F32));
+} // namespace
+
+TEST_SUITE(NEON)
+TEST_SUITE(ArithmeticSubtraction)
+
+template <typename T>
+using NEArithmeticSubtractionFixture = ArithmeticSubtractionValidationFixture<Tensor, Accessor, NEArithmeticSubtraction, T>;
+
+TEST_SUITE(U8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+               shape, policy)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::U8);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::U8);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::U8);
+
+    // Create and Configure function
+    NEArithmeticSubtraction sub;
+    sub.configure(&ref_src1, &ref_src2, &dst, policy);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(ref_src1.info()->padding(), padding);
+    validate(ref_src2.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEArithmeticSubtractionFixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionU8Dataset),
+                                                                                                                     framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(S16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::U8, DataType::S16 })),
+                                                                   framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+               shape, data_type, policy)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, data_type);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::S16);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::S16);
+
+    // Create and Configure function
+    NEArithmeticSubtraction sub;
+    sub.configure(&ref_src1, &ref_src2, &dst, policy);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(ref_src1.info()->padding(), padding);
+    validate(ref_src2.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEArithmeticSubtractionFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionS16Dataset),
+                                                                                                                     framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEArithmeticSubtractionFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(datasets::LargeShapes(), ArithmeticSubtractionS16Dataset),
+                                                                                                                   framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+template <typename T>
+using NEArithmeticSubtractionFixedPointFixture = ArithmeticSubtractionValidationFixedPointFixture<Tensor, Accessor, NEArithmeticSubtraction, T>;
+
+TEST_SUITE(Quantized)
+TEST_SUITE(QS8)
+FIXTURE_DATA_TEST_CASE(RunSmall, NEArithmeticSubtractionFixedPointFixture<int8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(), ArithmeticSubtractionQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 7)))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEArithmeticSubtractionFixedPointFixture<int8_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), ArithmeticSubtractionQS8Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 7)))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+
+TEST_SUITE(QS16)
+FIXTURE_DATA_TEST_CASE(RunSmall, NEArithmeticSubtractionFixedPointFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(datasets::SmallShapes(),
+                       ArithmeticSubtractionQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 15)))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEArithmeticSubtractionFixedPointFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(combine(combine(datasets::LargeShapes(), ArithmeticSubtractionQS16Dataset),
+                       framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+                       framework::dataset::make("FractionalBits", 1, 15)))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE(Float)
+#ifdef ARM_COMPUTE_ENABLE_FP16
+TEST_SUITE(FP16)
+FIXTURE_DATA_TEST_CASE(RunSmall, NEArithmeticSubtractionFixture<half_float::half>, framework::DatasetMode::ALL, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionFP16Dataset),
+                                                                                                                        framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+#endif /* ARM_COMPUTE_ENABLE_FP16 */
+
+TEST_SUITE(FP32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })),
+               shape, policy)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::F32);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::F32);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::F32);
+
+    // Create and Configure function
+    NEArithmeticSubtraction sub;
+    sub.configure(&ref_src1, &ref_src2, &dst, policy);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+
+    // Validate padding
+    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
+    validate(ref_src1.info()->padding(), padding);
+    validate(ref_src2.info()->padding(), padding);
+    validate(dst.info()->padding(), padding);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEArithmeticSubtractionFixture<float>, framework::DatasetMode::PRECOMMIT, combine(combine(datasets::SmallShapes(), ArithmeticSubtractionFP32Dataset),
+                                                                                                                   framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEArithmeticSubtractionFixture<float>, framework::DatasetMode::NIGHTLY, combine(combine(datasets::LargeShapes(), ArithmeticSubtractionFP32Dataset),
+                                                                                                                 framework::dataset::make("ConvertPolicy", { ConvertPolicy::SATURATE, ConvertPolicy::WRAP })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE_END()
+TEST_SUITE_END()
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/fixtures/ArithmeticAdditionFixture.h b/tests/validation/fixtures/ArithmeticAdditionFixture.h
index 5904ff385c..02c3ca9716 100644
--- a/tests/validation/fixtures/ArithmeticAdditionFixture.h
+++ b/tests/validation/fixtures/ArithmeticAdditionFixture.h
@@ -59,7 +59,7 @@ protected:
         library->fill_tensor_uniform(tensor, i);
     }
 
-    TensorType compute_target(const TensorShape &shape, DataType &data_type0, DataType &data_type1, DataType &output_data_type, ConvertPolicy &convert_policy, int fixed_point_position)
+    TensorType compute_target(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy, int fixed_point_position)
     {
         // Create tensors
         TensorType ref_src1 = create_tensor<TensorType>(shape, data_type0, 1, fixed_point_position);
@@ -93,7 +93,7 @@ protected:
         return dst;
     }
 
-    SimpleTensor<T> compute_reference(const TensorShape &shape, DataType &data_type0, DataType &data_type1, DataType &output_data_type, ConvertPolicy &convert_policy, int fixed_point_position)
+    SimpleTensor<T> compute_reference(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy, int fixed_point_position)
     {
         // Create reference
         SimpleTensor<T> ref_src1{ shape, data_type0, 1, fixed_point_position };
diff --git a/tests/validation/fixtures/ArithmeticSubtractionFixture.h b/tests/validation/fixtures/ArithmeticSubtractionFixture.h
new file mode 100644
index 0000000000..2c683d659f
--- /dev/null
+++ b/tests/validation/fixtures/ArithmeticSubtractionFixture.h
@@ -0,0 +1,126 @@
+/*
+ * 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_ARITHMETIC_SUBTRACTION_FIXTURE
+#define ARM_COMPUTE_TEST_ARITHMETIC_SUBTRACTION_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/CPP/ArithmeticSubtraction.h"
+#include "tests/validation/Helpers.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ArithmeticSubtractionValidationFixedPointFixture : public framework::Fixture
+{
+public:
+    template <typename...>
+    void setup(TensorShape shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy, int fractional_bits)
+    {
+        _fractional_bits = fractional_bits;
+        _target          = compute_target(shape, data_type0, data_type1, output_data_type, convert_policy, fractional_bits);
+        _reference       = compute_reference(shape, data_type0, data_type1, output_data_type, convert_policy, fractional_bits);
+    }
+
+protected:
+    template <typename U>
+    void fill(U &&tensor, int i)
+    {
+        library->fill_tensor_uniform(tensor, i);
+    }
+
+    TensorType compute_target(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy, int fixed_point_position)
+    {
+        // Create tensors
+        TensorType ref_src1 = create_tensor<TensorType>(shape, data_type0, 1, fixed_point_position);
+        TensorType ref_src2 = create_tensor<TensorType>(shape, data_type1, 1, fixed_point_position);
+        TensorType dst      = create_tensor<TensorType>(shape, output_data_type, 1, fixed_point_position);
+
+        // Create and configure function
+        FunctionType sub;
+        sub.configure(&ref_src1, &ref_src2, &dst, convert_policy);
+
+        ARM_COMPUTE_EXPECT(ref_src1.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(ref_src2.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(dst.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Allocate tensors
+        ref_src1.allocator()->allocate();
+        ref_src2.allocator()->allocate();
+        dst.allocator()->allocate();
+
+        ARM_COMPUTE_EXPECT(!ref_src1.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!ref_src2.info()->is_resizable(), framework::LogLevel::ERRORS);
+        ARM_COMPUTE_EXPECT(!dst.info()->is_resizable(), framework::LogLevel::ERRORS);
+
+        // Fill tensors
+        fill(AccessorType(ref_src1), 0);
+        fill(AccessorType(ref_src2), 1);
+
+        // Compute function
+        sub.run();
+
+        return dst;
+    }
+
+    SimpleTensor<T> compute_reference(const TensorShape &shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy, int fixed_point_position)
+    {
+        // Create reference
+        SimpleTensor<T> ref_src1{ shape, data_type0, 1, fixed_point_position };
+        SimpleTensor<T> ref_src2{ shape, data_type1, 1, fixed_point_position };
+
+        // Fill reference
+        fill(ref_src1, 0);
+        fill(ref_src2, 1);
+
+        return reference::arithmetic_subtraction<T>(ref_src1, ref_src2, output_data_type, convert_policy);
+    }
+
+    TensorType      _target{};
+    SimpleTensor<T> _reference{};
+    int             _fractional_bits{};
+};
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ArithmeticSubtractionValidationFixture : public ArithmeticSubtractionValidationFixedPointFixture<TensorType, AccessorType, FunctionType, T>
+{
+public:
+    template <typename...>
+    void setup(TensorShape shape, DataType data_type0, DataType data_type1, DataType output_data_type, ConvertPolicy convert_policy)
+    {
+        ArithmeticSubtractionValidationFixedPointFixture<TensorType, AccessorType, FunctionType, T>::setup(shape, data_type0, data_type1, output_data_type, convert_policy, 0);
+    }
+};
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
+#endif /* ARM_COMPUTE_TEST_ARITHMETIC_SUBTRACTION_FIXTURE */
diff --git a/tests/validation_old/CL/ArithmeticSubtraction.cpp b/tests/validation_old/CL/ArithmeticSubtraction.cpp
deleted file mode 100644
index 522ee7b1e8..0000000000
--- a/tests/validation_old/CL/ArithmeticSubtraction.cpp
+++ /dev/null
@@ -1,288 +0,0 @@
-/*
- * 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 "CL/CLAccessor.h"
-#include "PaddingCalculator.h"
-#include "TypePrinter.h"
-#include "tests/AssetsLibrary.h"
-#include "tests/Globals.h"
-#include "tests/Utils.h"
-#include "tests/validation_old/Datasets.h"
-#include "tests/validation_old/Reference.h"
-#include "tests/validation_old/Validation.h"
-
-#include "arm_compute/core/Helpers.h"
-#include "arm_compute/core/Types.h"
-#include "arm_compute/runtime/CL/functions/CLArithmeticSubtraction.h"
-#include "arm_compute/runtime/Tensor.h"
-#include "arm_compute/runtime/TensorAllocator.h"
-
-#include "tests/validation_old/boost_wrapper.h"
-
-#include <random>
-#include <string>
-
-using namespace arm_compute;
-using namespace arm_compute::test;
-using namespace arm_compute::test::validation;
-
-namespace
-{
-/** Compute Neon arithmetic subtraction function.
- *
- * @param[in] shape                Shape of the input and output tensors.
- * @param[in] dt_in0               Data type of first input tensor.
- * @param[in] dt_in1               Data type of second input tensor.
- * @param[in] dt_out               Data type of the output tensor.
- * @param[in] policy               Overflow policy of the operation.
- * @param[in] fixed_point_position (Optional) Fixed point position that expresses the number of bits for the fractional part of the number when the tensor's data type is QS8 or QS16 (default = 0).
- *
- * @return Computed output tensor.
- */
-CLTensor compute_arithmetic_subtraction(const TensorShape &shape, DataType dt_in0, DataType dt_in1, DataType dt_out, ConvertPolicy policy, int fixed_point_position = 0)
-{
-    // Create tensors
-    CLTensor src1 = create_tensor<CLTensor>(shape, dt_in0, 1, fixed_point_position);
-    CLTensor src2 = create_tensor<CLTensor>(shape, dt_in1, 1, fixed_point_position);
-    CLTensor dst  = create_tensor<CLTensor>(shape, dt_out, 1, fixed_point_position);
-
-    // Create and configure function
-    CLArithmeticSubtraction sub;
-    sub.configure(&src1, &src2, &dst, policy);
-
-    // Allocate tensors
-    src1.allocator()->allocate();
-    src2.allocator()->allocate();
-    dst.allocator()->allocate();
-
-    BOOST_TEST(!src1.info()->is_resizable());
-    BOOST_TEST(!src2.info()->is_resizable());
-    BOOST_TEST(!dst.info()->is_resizable());
-
-    // Fill tensors
-    library->fill_tensor_uniform(CLAccessor(src1), 0);
-    library->fill_tensor_uniform(CLAccessor(src2), 1);
-
-    // Compute function
-    sub.run();
-
-    return dst;
-}
-
-void validate_configuration(const CLTensor &src1, const CLTensor &src2, CLTensor &dst, TensorShape shape, ConvertPolicy policy)
-{
-    BOOST_TEST(src1.info()->is_resizable());
-    BOOST_TEST(src2.info()->is_resizable());
-    BOOST_TEST(dst.info()->is_resizable());
-
-    // Create and configure function
-    CLArithmeticSubtraction sub;
-    sub.configure(&src1, &src2, &dst, policy);
-
-    // Validate valid region
-    const ValidRegion valid_region = shape_to_valid_region(shape);
-    validate(src1.info()->valid_region(), valid_region);
-    validate(src2.info()->valid_region(), valid_region);
-    validate(dst.info()->valid_region(), valid_region);
-
-    // Validate padding
-    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
-    validate(src1.info()->padding(), padding);
-    validate(src2.info()->padding(), padding);
-    validate(dst.info()->padding(), padding);
-}
-} // namespace
-
-#ifndef DOXYGEN_SKIP_THIS
-BOOST_AUTO_TEST_SUITE(CL)
-BOOST_AUTO_TEST_SUITE(ArithmeticSubtraction)
-
-BOOST_AUTO_TEST_SUITE(U8)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit") * boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(Configuration, (SmallShapes() + LargeShapes()) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Create tensors
-    CLTensor src1 = create_tensor<CLTensor>(shape, DataType::U8);
-    CLTensor src2 = create_tensor<CLTensor>(shape, DataType::U8);
-    CLTensor dst  = create_tensor<CLTensor>(shape, DataType::U8);
-
-    validate_configuration(src1, src2, dst, shape, policy);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, DataType::U8, DataType::U8, DataType::U8, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::U8, DataType::U8, DataType::U8, policy);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE(S16)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit") * boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(Configuration, (SmallShapes() + LargeShapes()) * boost::unit_test::data::make({ DataType::U8, DataType::S16 }) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, dt, policy)
-{
-    // Create tensors
-    CLTensor src1 = create_tensor<CLTensor>(shape, dt);
-    CLTensor src2 = create_tensor<CLTensor>(shape, DataType::S16);
-    CLTensor dst  = create_tensor<CLTensor>(shape, DataType::S16);
-
-    validate_configuration(src1, src2, dst, shape, policy);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * boost::unit_test::data::make({ DataType::U8, DataType::S16 }) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, dt, policy)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * boost::unit_test::data::make({ DataType::U8, DataType::S16 }) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, dt, policy)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE(Quantized)
-BOOST_AUTO_TEST_SUITE(QS8)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 7),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 7),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE(QS16)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 15),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 15),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE(Float)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit") * boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(Configuration, (SmallShapes() + LargeShapes()) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Create tensors
-    CLTensor src1 = create_tensor<CLTensor>(shape, DataType::F32);
-    CLTensor src2 = create_tensor<CLTensor>(shape, DataType::F32);
-    CLTensor dst  = create_tensor<CLTensor>(shape, DataType::F32);
-
-    validate_configuration(src1, src2, dst, shape, policy);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes(), shape)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, ConvertPolicy::WRAP);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, ConvertPolicy::WRAP);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Compute function
-    CLTensor dst = compute_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, policy);
-
-    // Validate output
-    validate(CLAccessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE_END()
-BOOST_AUTO_TEST_SUITE_END()
-#endif /* DOXYGEN_SKIP_THIS */
diff --git a/tests/validation_old/NEON/ArithmeticSubtraction.cpp b/tests/validation_old/NEON/ArithmeticSubtraction.cpp
deleted file mode 100644
index 86aa124f00..0000000000
--- a/tests/validation_old/NEON/ArithmeticSubtraction.cpp
+++ /dev/null
@@ -1,306 +0,0 @@
-/*
- * 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 "NEON/Accessor.h"
-#include "PaddingCalculator.h"
-#include "TypePrinter.h"
-#include "Utils.h"
-#include "tests/AssetsLibrary.h"
-#include "tests/Globals.h"
-#include "tests/validation_old/Datasets.h"
-#include "tests/validation_old/Reference.h"
-#include "tests/validation_old/Validation.h"
-
-#include "arm_compute/core/Helpers.h"
-#include "arm_compute/core/Types.h"
-#include "arm_compute/runtime/NEON/functions/NEArithmeticSubtraction.h"
-#include "arm_compute/runtime/Tensor.h"
-#include "arm_compute/runtime/TensorAllocator.h"
-
-#include "tests/validation_old/boost_wrapper.h"
-
-#include <random>
-#include <string>
-
-using namespace arm_compute;
-using namespace arm_compute::test;
-using namespace arm_compute::test::validation;
-
-namespace
-{
-/** Compute Neon arithmetic subtraction function.
- *
- * @param[in] shape                Shape of the input and output tensors.
- * @param[in] dt_in0               Data type of first input tensor.
- * @param[in] dt_in1               Data type of second input tensor.
- * @param[in] dt_out               Data type of the output tensor.
- * @param[in] policy               Overflow policy of the operation.
- * @param[in] fixed_point_position (Optional) Fixed point position that expresses the number of bits for the fractional part of the number when the tensor's data type is QS8 or QS16 (default = 0).
- *
- * @return Computed output tensor.
- */
-Tensor compute_arithmetic_subtraction(const TensorShape &shape, DataType dt_in0, DataType dt_in1, DataType dt_out, ConvertPolicy policy, int fixed_point_position = 0)
-{
-    // Create tensors
-    Tensor src1 = create_tensor<Tensor>(shape, dt_in0, 1, fixed_point_position);
-    Tensor src2 = create_tensor<Tensor>(shape, dt_in1, 1, fixed_point_position);
-    Tensor dst  = create_tensor<Tensor>(shape, dt_out, 1, fixed_point_position);
-
-    // Create and configure function
-    NEArithmeticSubtraction sub;
-    sub.configure(&src1, &src2, &dst, policy);
-
-    // Allocate tensors
-    src1.allocator()->allocate();
-    src2.allocator()->allocate();
-    dst.allocator()->allocate();
-
-    BOOST_TEST(!src1.info()->is_resizable());
-    BOOST_TEST(!src2.info()->is_resizable());
-    BOOST_TEST(!dst.info()->is_resizable());
-
-    // Fill tensors
-    library->fill_tensor_uniform(Accessor(src1), 0);
-    library->fill_tensor_uniform(Accessor(src2), 1);
-
-    // Compute function
-    sub.run();
-
-    return dst;
-}
-
-void validate_configuration(const Tensor &src1, const Tensor &src2, Tensor &dst, TensorShape shape, ConvertPolicy policy)
-{
-    BOOST_TEST(src1.info()->is_resizable());
-    BOOST_TEST(src2.info()->is_resizable());
-    BOOST_TEST(dst.info()->is_resizable());
-
-    // Create and configure function
-    NEArithmeticSubtraction sub;
-    sub.configure(&src1, &src2, &dst, policy);
-
-    // Validate valid region
-    const ValidRegion valid_region = shape_to_valid_region(shape);
-    validate(src1.info()->valid_region(), valid_region);
-    validate(src2.info()->valid_region(), valid_region);
-    validate(dst.info()->valid_region(), valid_region);
-
-    // Validate padding
-    const PaddingSize padding = PaddingCalculator(shape.x(), 16).required_padding();
-    validate(src1.info()->padding(), padding);
-    validate(src2.info()->padding(), padding);
-    validate(dst.info()->padding(), padding);
-}
-} // namespace
-
-#ifndef DOXYGEN_SKIP_THIS
-BOOST_AUTO_TEST_SUITE(NEON)
-BOOST_AUTO_TEST_SUITE(ArithmeticSubtraction)
-
-BOOST_AUTO_TEST_SUITE(U8)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit") * boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(Configuration, (SmallShapes() + LargeShapes()) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Create tensors
-    Tensor src1 = create_tensor<Tensor>(shape, DataType::U8);
-    Tensor src2 = create_tensor<Tensor>(shape, DataType::U8);
-    Tensor dst  = create_tensor<Tensor>(shape, DataType::U8);
-
-    validate_configuration(src1, src2, dst, shape, policy);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::U8, DataType::U8, DataType::U8, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::U8, DataType::U8, DataType::U8, policy);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE(S16)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit") * boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(Configuration, (SmallShapes() + LargeShapes()) * boost::unit_test::data::make({ DataType::U8, DataType::S16 }) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, dt, policy)
-{
-    // Create tensors
-    Tensor src1 = create_tensor<Tensor>(shape, dt);
-    Tensor src2 = create_tensor<Tensor>(shape, DataType::S16);
-    Tensor dst  = create_tensor<Tensor>(shape, DataType::S16);
-
-    validate_configuration(src1, src2, dst, shape, policy);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * boost::unit_test::data::make({ DataType::U8, DataType::S16 }) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, dt, policy)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * boost::unit_test::data::make({ DataType::U8, DataType::S16 }) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, dt, policy)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, dt, DataType::S16, DataType::S16, policy);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE(Quantized)
-BOOST_AUTO_TEST_SUITE(QS8)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 7),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 7),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS8, DataType::QS8, DataType::QS8, policy, fixed_point_position);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE(QS16)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 15),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * ConvertPolicies() * boost::unit_test::data::xrange(1, 15),
-                     shape, policy, fixed_point_position)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::QS16, DataType::QS16, DataType::QS16, policy, fixed_point_position);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-BOOST_AUTO_TEST_SUITE_END()
-
-#ifdef ARM_COMPUTE_ENABLE_FP16
-BOOST_AUTO_TEST_SUITE(Float16)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes(), shape)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::F16, DataType::F16, DataType::F16, ConvertPolicy::WRAP);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::F16, DataType::F16, DataType::F16, ConvertPolicy::WRAP);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-#endif /* ARM_COMPUTE_ENABLE_FP16 */
-
-BOOST_AUTO_TEST_SUITE(Float)
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit") * boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(Configuration, (SmallShapes() + LargeShapes()) * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Create tensors
-    Tensor src1 = create_tensor<Tensor>(shape, DataType::F32);
-    Tensor src2 = create_tensor<Tensor>(shape, DataType::F32);
-    Tensor dst  = create_tensor<Tensor>(shape, DataType::F32);
-
-    validate_configuration(src1, src2, dst, shape, policy);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("precommit"))
-BOOST_DATA_TEST_CASE(RunSmall, SmallShapes(), shape)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, ConvertPolicy::WRAP);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, ConvertPolicy::WRAP);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_TEST_DECORATOR(*boost::unit_test::label("nightly"))
-BOOST_DATA_TEST_CASE(RunLarge, LargeShapes() * boost::unit_test::data::make({ ConvertPolicy::SATURATE, ConvertPolicy::WRAP }),
-                     shape, policy)
-{
-    // Compute function
-    Tensor dst = compute_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, policy);
-
-    // Compute reference
-    RawTensor ref_dst = Reference::compute_reference_arithmetic_subtraction(shape, DataType::F32, DataType::F32, DataType::F32, policy);
-
-    // Validate output
-    validate(Accessor(dst), ref_dst);
-}
-BOOST_AUTO_TEST_SUITE_END()
-
-BOOST_AUTO_TEST_SUITE_END()
-BOOST_AUTO_TEST_SUITE_END()
-#endif /* DOXYGEN_SKIP_THIS */
diff --git a/tests/validation_old/Reference.cpp b/tests/validation_old/Reference.cpp
index f8f2ded1dc..5aa39cb22e 100644
--- a/tests/validation_old/Reference.cpp
+++ b/tests/validation_old/Reference.cpp
@@ -187,23 +187,6 @@ RawTensor Reference::compute_reference_accumulate_weighted(const TensorShape &sh
     return ref_dst;
 }
 
-RawTensor Reference::compute_reference_arithmetic_subtraction(const TensorShape &shape, DataType dt_in0, DataType dt_in1, DataType dt_out, ConvertPolicy convert_policy, int fixed_point_position)
-{
-    // Create reference
-    RawTensor ref_src1(shape, dt_in0, 1, fixed_point_position);
-    RawTensor ref_src2(shape, dt_in1, 1, fixed_point_position);
-    RawTensor ref_dst(shape, dt_out, 1, fixed_point_position);
-
-    // Fill reference
-    library->fill_tensor_uniform(ref_src1, 0);
-    library->fill_tensor_uniform(ref_src2, 1);
-
-    // Compute reference
-    ReferenceCPP::arithmetic_subtraction(ref_src1, ref_src2, ref_dst, convert_policy);
-
-    return ref_dst;
-}
-
 RawTensor Reference::compute_reference_depth_convert(const TensorShape &shape, DataType dt_in, DataType dt_out, ConvertPolicy policy,
                                                      uint32_t shift, uint32_t fixed_point_position_in, uint32_t fixed_point_position_out)
 {
diff --git a/tests/validation_old/Reference.h b/tests/validation_old/Reference.h
index 3be7de1b62..60383a4576 100644
--- a/tests/validation_old/Reference.h
+++ b/tests/validation_old/Reference.h
@@ -130,18 +130,6 @@ public:
      * @return Computed raw tensor.
      */
     static RawTensor compute_reference_accumulate_weighted(const TensorShape &shape, float alpha);
-    /** Compute reference arithmetic subtraction.
-     *
-     * @param[in] shape                Shape of the input and output tensors.
-     * @param[in] dt_in0               Data type of first input tensor.
-     * @param[in] dt_in1               Data type of second input tensor.
-     * @param[in] dt_out               Data type of the output tensor.
-     * @param[in] convert_policy       Overflow policy of the operation.
-     * @param[in] fixed_point_position (Optional) Number of bits for the fractional part of the fixed point numbers
-     *
-     * @return Computed raw tensor.
-     */
-    static RawTensor compute_reference_arithmetic_subtraction(const TensorShape &shape, DataType dt_in0, DataType dt_in1, DataType dt_out, ConvertPolicy convert_policy, int fixed_point_position = 0);
     /** Compute reference depth convert.
      *
      * @param[in] shape                    Shape of the input and output tensors.
diff --git a/tests/validation_old/ReferenceCPP.cpp b/tests/validation_old/ReferenceCPP.cpp
index 9c7ddf8579..0247b702ff 100644
--- a/tests/validation_old/ReferenceCPP.cpp
+++ b/tests/validation_old/ReferenceCPP.cpp
@@ -147,15 +147,6 @@ void ReferenceCPP::accumulate_weighted(const RawTensor &src, RawTensor &dst, flo
     tensor_operations::accumulate_weighted(s, d, alpha);
 }
 
-// Arithmetic subtraction
-void ReferenceCPP::arithmetic_subtraction(const RawTensor &src1, const RawTensor &src2, RawTensor &dst, ConvertPolicy convert_policy)
-{
-    const TensorVariant s1 = TensorFactory::get_tensor(src1);
-    const TensorVariant s2 = TensorFactory::get_tensor(src2);
-    TensorVariant       d  = TensorFactory::get_tensor(dst);
-    boost::apply_visitor(arithmetic_subtraction_visitor(convert_policy), s1, s2, d);
-}
-
 // Depth conversion
 void ReferenceCPP::depth_convert(const RawTensor &src, RawTensor &dst, ConvertPolicy policy, uint32_t shift)
 {
diff --git a/tests/validation_old/ReferenceCPP.h b/tests/validation_old/ReferenceCPP.h
index 611e98a780..2ac1e5784b 100644
--- a/tests/validation_old/ReferenceCPP.h
+++ b/tests/validation_old/ReferenceCPP.h
@@ -127,14 +127,6 @@ public:
      * @param[in]      alpha A float value within the range of [0, 1]
      */
     static void accumulate_weighted(const RawTensor &src, RawTensor &dst, float alpha);
-    /** Arithmetic subtraction of @p src2 from @p src1
-     *
-     * @param[in]  src1           First tensor.
-     * @param[in]  src2           Second tensor.
-     * @param[out] dst            Result tensor.
-     * @param[in]  convert_policy Overflow policy.
-     */
-    static void arithmetic_subtraction(const RawTensor &src1, const RawTensor &src2, RawTensor &dst, ConvertPolicy convert_policy);
     /** Depth conversion from @p src to @p dst
      *
      * @param[in]  src    First tensor.
diff --git a/tests/validation_old/TensorOperations.h b/tests/validation_old/TensorOperations.h
index 7f29788fc7..79d3720b84 100644
--- a/tests/validation_old/TensorOperations.h
+++ b/tests/validation_old/TensorOperations.h
@@ -577,19 +577,6 @@ void accumulate_weighted(const Tensor<T> &in, Tensor<T> &out, float alpha)
     }
 }
 
-// Arithmetic Subtraction
-template <typename T1, typename T2, typename T3>
-void arithmetic_subtraction(const Tensor<T1> &in1, const Tensor<T2> &in2, Tensor<T3> &out, ConvertPolicy convert_policy)
-{
-    using intermediate_type = typename common_promoted_signed_type<T1, T2, T3>::intermediate_type;
-
-    for(int i = 0; i < in1.num_elements(); ++i)
-    {
-        intermediate_type val = static_cast<intermediate_type>(in1[i]) - static_cast<intermediate_type>(in2[i]);
-        out[i]                = (convert_policy == ConvertPolicy::SATURATE) ? saturate_cast<T3>(val) : static_cast<T3>(val);
-    }
-}
-
 // Depth conversion
 template < typename T1, typename T2, typename std::enable_if < std::is_integral<T1>::value &&is_floating_point<T2>::value, int >::type = 0 >
 void depth_convert(const Tensor<T1> &in, Tensor<T2> &out, ConvertPolicy policy, uint32_t shift)
diff --git a/tests/validation_old/TensorVisitors.h b/tests/validation_old/TensorVisitors.h
index 21ec11181d..ead7f389d5 100644
--- a/tests/validation_old/TensorVisitors.h
+++ b/tests/validation_old/TensorVisitors.h
@@ -78,24 +78,6 @@ public:
         tensor_operations::absolute_difference(in1, in2, out);
     }
 };
-// Arithmetic Subtraction visitor
-struct arithmetic_subtraction_visitor : public boost::static_visitor<>
-{
-public:
-    explicit arithmetic_subtraction_visitor(ConvertPolicy convert_policy)
-        : _policy(convert_policy)
-    {
-    }
-
-    template <typename T1, typename T2, typename T3>
-    void operator()(const Tensor<T1> &in1, const Tensor<T2> &in2, Tensor<T3> &out) const
-    {
-        tensor_operations::arithmetic_subtraction(in1, in2, out, _policy);
-    }
-
-private:
-    ConvertPolicy _policy;
-};
 // Depth Convert visitor
 struct depth_convert_visitor : public boost::static_visitor<>
 {