COMPMID-1754: NEON: Implement Maximum, Minumum, SquaredDifference

Change-Id: I77e8c6a8af6ad841293ed5e66ed582035cc1424b
Reviewed-on: https://review.mlplatform.org/339
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>
Reviewed-by: Georgios Pinitas <georgios.pinitas@arm.com>

5 files changed, 846 insertions, 0 deletions

diff --git a/tests/validation/NEON/ElementwiseMax.cpp b/tests/validation/NEON/ElementwiseMax.cpp
new file mode 100644
index 0000000000..c77f485d29
--- /dev/null
+++ b/tests/validation/NEON/ElementwiseMax.cpp
@@ -0,0 +1,264 @@
+/*
+ * Copyright (c) 2018 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/NEElementwiseOperations.h"
+#include "arm_compute/runtime/Tensor.h"
+#include "arm_compute/runtime/TensorAllocator.h"
+#include "tests/NEON/Accessor.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/ElementwiseOperationsFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+RelativeTolerance<float> tolerance_fp32(0.000001f);
+/** Input data sets **/
+const auto ElementwiseMaxQASYMM8Dataset = combine(combine(framework::dataset::make("DataType", DataType::QASYMM8), framework::dataset::make("DataType", DataType::QASYMM8)),
+                                                  framework::dataset::make("DataType",
+                                                                           DataType::QASYMM8));
+/** Input data sets **/
+const auto ElementwiseMaxS32Dataset = combine(combine(framework::dataset::make("DataType", DataType::S32), framework::dataset::make("DataType", DataType::S32)), framework::dataset::make("DataType",
+                                              DataType::S32));
+const auto ElementwiseMaxS16Dataset = combine(combine(framework::dataset::make("DataType", { DataType::S16 }), framework::dataset::make("DataType", DataType::S16)),
+                                              framework::dataset::make("DataType", DataType::S16));
+#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+const auto ElementwiseMaxFP16Dataset = combine(combine(framework::dataset::make("DataType", DataType::F16), framework::dataset::make("DataType", DataType::F16)),
+                                               framework::dataset::make("DataType", DataType::F16));
+#endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
+const auto ElementwiseMaxFP32Dataset = 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(ElementwiseMax)
+
+template <typename T>
+using NEElementwiseMaxFixture = ElementwiseMaxValidationFixture<Tensor, Accessor, NEElementwiseMax, T>;
+
+// *INDENT-OFF*
+// clang-format off
+DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, zip(zip(zip(
+               framework::dataset::make("Input1Info", { TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                        TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),      // Invalid data type combination
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),     // Mismatching shapes
+                                                      }),
+               framework::dataset::make("Input2Info",{ TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S16),
+                                                       TensorInfo(TensorShape(48U, 11U, 2U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("OutputInfo",{ TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(48U, 11U, 2U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("Expected", { true, true, true, false, false})),
+               input1_info, input2_info, output_info, expected)
+{
+    ARM_COMPUTE_EXPECT(bool(NEElementwiseMax::validate(&input1_info.clone()->set_is_resizable(false), &input2_info.clone()->set_is_resizable(false), &output_info.clone()->set_is_resizable(false))) == expected, framework::LogLevel::ERRORS);
+}
+// clang-format on
+// *INDENT-ON*
+
+TEST_SUITE(S32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::S32);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::S32);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::S32);
+
+    // Create and Configure function
+    NEElementwiseMax add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMaxFixture<int32_t>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseMaxS32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // S32
+
+TEST_SUITE(S16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::S16 })),
+               shape, data_type)
+{
+    // 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
+    NEElementwiseMax add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMaxFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseMaxS16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEElementwiseMaxFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), ElementwiseMaxS16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // S16
+
+template <typename T>
+using NEElementwiseMaxQuantizedFixture = ElementwiseMaxValidationQuantizedFixture<Tensor, Accessor, NEElementwiseMax, T>;
+
+TEST_SUITE(Quantized)
+TEST_SUITE(QASYMM8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::QASYMM8);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::QASYMM8);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::QASYMM8);
+
+    // Create and Configure function
+    NEElementwiseMin add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMaxQuantizedFixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(combine(datasets::SmallShapes(),
+                                                                                                                       ElementwiseMaxQASYMM8Dataset),
+                                                                                                                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(5.f / 255.f, 20) })),
+                                                                                                                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(2.f / 255.f, 10) })),
+                                                                                                                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(1.f / 255.f, 5) })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference, tolerance_fp32, 0.01);
+}
+
+template <typename T>
+using NEElementwiseMaxQuantizedBroadcastFixture = ElementwiseMaxQuantizedBroadcastValidationFixture<Tensor, Accessor, NEElementwiseMax, T>;
+
+FIXTURE_DATA_TEST_CASE(RunSmallBroadcast, NEElementwiseMaxQuantizedBroadcastFixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(combine(datasets::SmallShapesBroadcast(),
+                       ElementwiseMaxQASYMM8Dataset),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(5.f / 255.f, 20) })),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(2.f / 255.f, 10) })),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(1.f / 255.f, 5) })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE(Float)
+#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+TEST_SUITE(F16)
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMaxFixture<half>, framework::DatasetMode::ALL, combine(datasets::SmallShapes(), ElementwiseMaxFP16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // F16
+#endif           /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
+
+TEST_SUITE(F32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // 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
+    NEElementwiseMax add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMaxFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseMaxFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEElementwiseMaxFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), ElementwiseMaxFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+template <typename T>
+using NEElementwiseMaxBroadcastFixture = ElementwiseMaxBroadcastValidationFixture<Tensor, Accessor, NEElementwiseMax, T>;
+
+FIXTURE_DATA_TEST_CASE(RunSmallBroadcast, NEElementwiseMaxBroadcastFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapesBroadcast(),
+                       ElementwiseMaxFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLargeBroadcast, NEElementwiseMaxBroadcastFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapesBroadcast(),
+                       ElementwiseMaxFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // F32
+TEST_SUITE_END() // Float
+
+TEST_SUITE_END() // ElementwiseMax
+TEST_SUITE_END() // NEON
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/NEON/ElementwiseMin.cpp b/tests/validation/NEON/ElementwiseMin.cpp
new file mode 100644
index 0000000000..a7ce2d61b0
--- /dev/null
+++ b/tests/validation/NEON/ElementwiseMin.cpp
@@ -0,0 +1,265 @@
+/*
+ * Copyright (c) 2018 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/NEElementwiseOperations.h"
+#include "arm_compute/runtime/Tensor.h"
+#include "arm_compute/runtime/TensorAllocator.h"
+#include "tests/NEON/Accessor.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/ElementwiseOperationsFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+RelativeTolerance<float> tolerance_fp32(0.000001f);
+/** Input data sets **/
+const auto ElementwiseMinQASYMM8Dataset = combine(combine(framework::dataset::make("DataType", DataType::QASYMM8), framework::dataset::make("DataType", DataType::QASYMM8)),
+                                                  framework::dataset::make("DataType",
+                                                                           DataType::QASYMM8));
+const auto ElementwiseMinS32Dataset = combine(combine(framework::dataset::make("DataType", DataType::S32), framework::dataset::make("DataType", DataType::S32)), framework::dataset::make("DataType",
+                                              DataType::S32));
+const auto ElementwiseMinS16Dataset = combine(combine(framework::dataset::make("DataType", { DataType::S16 }), framework::dataset::make("DataType", DataType::S16)),
+                                              framework::dataset::make("DataType", DataType::S16));
+#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+const auto ElementwiseMinFP16Dataset = combine(combine(framework::dataset::make("DataType", DataType::F16), framework::dataset::make("DataType", DataType::F16)),
+                                               framework::dataset::make("DataType", DataType::F16));
+#endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
+const auto ElementwiseMinFP32Dataset = 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(ElementwiseMin)
+
+template <typename T>
+using NEElementwiseMinFixture = ElementwiseMinValidationFixture<Tensor, Accessor, NEElementwiseMin, T>;
+
+// *INDENT-OFF*
+// clang-format off
+DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, zip(zip(zip(
+               framework::dataset::make("Input1Info", { TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                        TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),      // Invalid data type combination
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),     // Mismatching shapes
+                                                      }),
+               framework::dataset::make("Input2Info",{ TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S16),
+                                                       TensorInfo(TensorShape(48U, 11U, 2U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("OutputInfo",{ TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(48U, 11U, 2U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("Expected", { true, true, true, false, false})),
+               input1_info, input2_info, output_info, expected)
+{
+    ARM_COMPUTE_EXPECT(bool(NEElementwiseMin::validate(&input1_info.clone()->set_is_resizable(false), &input2_info.clone()->set_is_resizable(false), &output_info.clone()->set_is_resizable(false))) == expected, framework::LogLevel::ERRORS);
+}
+// clang-format on
+// *INDENT-ON*
+
+TEST_SUITE(S32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::S32);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::S32);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::S32);
+
+    // Create and Configure function
+    NEElementwiseMin add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMinFixture<int32_t>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseMinS32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // S32
+
+TEST_SUITE(S16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::S16 })),
+               shape, data_type)
+{
+    // 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
+    NEElementwiseMin add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMinFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseMinS16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEElementwiseMinFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), ElementwiseMinS16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // S16
+
+template <typename T>
+using NEElementwiseMinQuantizedFixture = ElementwiseMinValidationQuantizedFixture<Tensor, Accessor, NEElementwiseMin, T>;
+
+TEST_SUITE(Quantized)
+TEST_SUITE(QASYMM8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::QASYMM8);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::QASYMM8);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::QASYMM8);
+
+    // Create and Configure function
+    NEElementwiseMin add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+template <typename T>
+using NEElementwiseMinQuantizedBroadcastFixture = ElementwiseMinQuantizedBroadcastValidationFixture<Tensor, Accessor, NEElementwiseMin, T>;
+
+FIXTURE_DATA_TEST_CASE(RunSmallBroadcast, NEElementwiseMinQuantizedBroadcastFixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(combine(datasets::SmallShapesBroadcast(),
+                       ElementwiseMinQASYMM8Dataset),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(5.f / 255.f, 20) })),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(2.f / 255.f, 10) })),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(1.f / 255.f, 5) })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMinQuantizedFixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(combine(datasets::SmallShapes(),
+                                                                                                                       ElementwiseMinQASYMM8Dataset),
+                                                                                                                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(5.f / 255.f, 20) })),
+                                                                                                                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(2.f / 255.f, 10) })),
+                                                                                                                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(1.f / 255.f, 5) }))
+
+                      )
+{
+    // Validate output
+    validate(Accessor(_target), _reference, tolerance_fp32, 0.01);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE(Float)
+#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+TEST_SUITE(F16)
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMinFixture<half>, framework::DatasetMode::ALL, combine(datasets::SmallShapes(), ElementwiseMinFP16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // F16
+#endif           /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
+
+TEST_SUITE(F32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // 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
+    NEElementwiseMin add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseMinFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseMinFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEElementwiseMinFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), ElementwiseMinFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+template <typename T>
+using NEElementwiseMinBroadcastFixture = ElementwiseMinBroadcastValidationFixture<Tensor, Accessor, NEElementwiseMin, T>;
+
+FIXTURE_DATA_TEST_CASE(RunSmallBroadcast, NEElementwiseMinBroadcastFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapesBroadcast(),
+                       ElementwiseMinFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLargeBroadcast, NEElementwiseMinBroadcastFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapesBroadcast(),
+                       ElementwiseMinFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // F32
+TEST_SUITE_END() // Float
+
+TEST_SUITE_END() // ElementwiseMin
+TEST_SUITE_END() // NEON
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/NEON/ElementwiseSquareDiff.cpp b/tests/validation/NEON/ElementwiseSquareDiff.cpp
new file mode 100644
index 0000000000..0c3fab609e
--- /dev/null
+++ b/tests/validation/NEON/ElementwiseSquareDiff.cpp
@@ -0,0 +1,268 @@
+/*
+ * Copyright (c) 2018 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/NEElementwiseOperations.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/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/ElementwiseOperationsFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+namespace
+{
+RelativeTolerance<float> tolerance_fp32(0.000001f);
+/** Input data sets **/
+const auto ElementwiseSquaredDiffQASYMM8Dataset = combine(combine(framework::dataset::make("DataType", DataType::QASYMM8), framework::dataset::make("DataType", DataType::QASYMM8)),
+                                                          framework::dataset::make("DataType",
+                                                                                   DataType::QASYMM8));
+/** Input data sets **/
+const auto ElementwiseSquaredDiffS32Dataset = combine(combine(framework::dataset::make("DataType", DataType::S32), framework::dataset::make("DataType", DataType::S32)),
+                                                      framework::dataset::make("DataType",
+                                                                               DataType::S32));
+const auto ElementwiseSquaredDiffS16Dataset = combine(combine(framework::dataset::make("DataType", { DataType::S16 }), framework::dataset::make("DataType", DataType::S16)),
+                                                      framework::dataset::make("DataType", DataType::S16));
+#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+const auto ElementwiseSquaredDiffFP16Dataset = combine(combine(framework::dataset::make("DataType", DataType::F16), framework::dataset::make("DataType", DataType::F16)),
+                                                       framework::dataset::make("DataType", DataType::F16));
+#endif /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
+const auto ElementwiseSquaredDiffFP32Dataset = 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(ElementwiseSquaredDiff)
+
+template <typename T>
+using NEElementwiseSquaredDiffFixture = ElementwiseSquaredDiffValidationFixture<Tensor, Accessor, NEElementwiseSquaredDiff, T>;
+
+// *INDENT-OFF*
+// clang-format off
+DATA_TEST_CASE(Validate, framework::DatasetMode::ALL, zip(zip(zip(
+               framework::dataset::make("Input1Info", { TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                        TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),      // Invalid data type combination
+                                                        TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),     // Mismatching shapes
+                                                      }),
+               framework::dataset::make("Input2Info",{ TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S16),
+                                                       TensorInfo(TensorShape(48U, 11U, 2U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("OutputInfo",{ TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::F32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(27U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(32U, 13U, 2U), 1, DataType::S32),
+                                                       TensorInfo(TensorShape(48U, 11U, 2U), 1, DataType::F32),
+                                                     })),
+               framework::dataset::make("Expected", { true, true, true, false, false})),
+               input1_info, input2_info, output_info, expected)
+{
+    ARM_COMPUTE_EXPECT(bool(NEElementwiseSquaredDiff::validate(&input1_info.clone()->set_is_resizable(false), &input2_info.clone()->set_is_resizable(false), &output_info.clone()->set_is_resizable(false))) == expected, framework::LogLevel::ERRORS);
+}
+// clang-format on
+// *INDENT-ON*
+
+TEST_SUITE(S32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::S32);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::S32);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::S32);
+
+    // Create and Configure function
+    NEElementwiseSquaredDiff add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseSquaredDiffFixture<int32_t>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseSquaredDiffS32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // S32
+
+TEST_SUITE(S16)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, combine(framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()), framework::dataset::make("DataType", { DataType::S16 })),
+               shape, data_type)
+{
+    // 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
+    NEElementwiseSquaredDiff add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseSquaredDiffFixture<int16_t>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseSquaredDiffS16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEElementwiseSquaredDiffFixture<int16_t>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), ElementwiseSquaredDiffS16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // S16
+
+template <typename T>
+using NEElementwiseSquaredDiffQuantizedFixture = ElementwiseSquaredDiffValidationQuantizedFixture<Tensor, Accessor, NEElementwiseSquaredDiff, T>;
+
+TEST_SUITE(Quantized)
+TEST_SUITE(QASYMM8)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // Create tensors
+    Tensor ref_src1 = create_tensor<Tensor>(shape, DataType::QASYMM8);
+    Tensor ref_src2 = create_tensor<Tensor>(shape, DataType::QASYMM8);
+    Tensor dst      = create_tensor<Tensor>(shape, DataType::QASYMM8);
+
+    // Create and Configure function
+    NEElementwiseMin add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseSquaredDiffQuantizedFixture<uint8_t>, framework::DatasetMode::PRECOMMIT, combine(combine(combine(combine(datasets::SmallShapes(),
+                       ElementwiseSquaredDiffQASYMM8Dataset),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(5.f / 255.f, 20) })),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(2.f / 255.f, 10) })),
+                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(1.f / 255.f, 5) }))
+
+                      )
+{
+    // Validate output
+    validate(Accessor(_target), _reference, tolerance_fp32, 0.01);
+}
+template <typename T>
+using NEElementwiseSquaredDiffQuantizedBroadcastFixture = ElementwiseSquaredDiffQuantizedBroadcastValidationFixture<Tensor, Accessor, NEElementwiseSquaredDiff, T>;
+
+FIXTURE_DATA_TEST_CASE(RunSmallBroadcast, NEElementwiseSquaredDiffQuantizedBroadcastFixture<uint8_t>, framework::DatasetMode::PRECOMMIT,
+                       combine(combine(combine(combine(datasets::SmallShapesBroadcast(),
+                                                       ElementwiseSquaredDiffQASYMM8Dataset),
+                                               framework::dataset::make("QuantizationInfo", { QuantizationInfo(5.f / 255.f, 20) })),
+                                       framework::dataset::make("QuantizationInfo", { QuantizationInfo(2.f / 255.f, 10) })),
+                               framework::dataset::make("QuantizationInfo", { QuantizationInfo(1.f / 255.f, 5) })))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END()
+TEST_SUITE_END()
+
+TEST_SUITE(Float)
+#ifdef __ARM_FEATURE_FP16_VECTOR_ARITHMETIC
+TEST_SUITE(F16)
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseSquaredDiffFixture<half>, framework::DatasetMode::ALL, combine(datasets::SmallShapes(), ElementwiseSquaredDiffFP16Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // F16
+#endif           /* __ARM_FEATURE_FP16_VECTOR_ARITHMETIC */
+
+TEST_SUITE(F32)
+DATA_TEST_CASE(Configuration, framework::DatasetMode::ALL, framework::dataset::concat(datasets::SmallShapes(), datasets::LargeShapes()),
+               shape)
+{
+    // 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
+    NEElementwiseSquaredDiff add;
+    add.configure(&ref_src1, &ref_src2, &dst);
+
+    // Validate valid region
+    const ValidRegion valid_region = shape_to_valid_region(shape);
+    validate(dst.info()->valid_region(), valid_region);
+}
+
+FIXTURE_DATA_TEST_CASE(RunSmall, NEElementwiseSquaredDiffFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), ElementwiseSquaredDiffFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLarge, NEElementwiseSquaredDiffFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), ElementwiseSquaredDiffFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+template <typename T>
+using NEElementwiseSquaredDiffBroadcastFixture = ElementwiseSquaredDiffBroadcastValidationFixture<Tensor, Accessor, NEElementwiseSquaredDiff, T>;
+
+FIXTURE_DATA_TEST_CASE(RunSmallBroadcast, NEElementwiseSquaredDiffBroadcastFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapesBroadcast(),
+                       ElementwiseSquaredDiffFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+
+FIXTURE_DATA_TEST_CASE(RunLargeBroadcast, NEElementwiseSquaredDiffBroadcastFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapesBroadcast(),
+                       ElementwiseSquaredDiffFP32Dataset))
+{
+    // Validate output
+    validate(Accessor(_target), _reference);
+}
+TEST_SUITE_END() // F32
+TEST_SUITE_END() // Float
+
+TEST_SUITE_END() // ElementwiseSquaredDiff
+TEST_SUITE_END() // NEON
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/tests/validation/fixtures/ElementwiseOperationsFixture.h b/tests/validation/fixtures/ElementwiseOperationsFixture.h
index b051c858c2..8190b2405a 100644
--- a/tests/validation/fixtures/ElementwiseOperationsFixture.h
+++ b/tests/validation/fixtures/ElementwiseOperationsFixture.h
@@ -200,6 +200,21 @@ public:
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ElementwiseMaxQuantizedBroadcastValidationFixture : public ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>
+{
+public:
+    template <typename...>
+    void setup(const TensorShape &shape0, const TensorShape &shape1, DataType data_type0, DataType data_type1, DataType output_data_type,
+               QuantizationInfo qinfo0, QuantizationInfo qinfo1, QuantizationInfo qinfo_out)
+
+    {
+        ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(ArithmeticOperation::MAX, shape0, shape1,
+                                                                                             data_type0, data_type1, output_data_type,
+                                                                                             qinfo0, qinfo1, qinfo_out);
+    }
+};
+
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
 class ElementwiseMinBroadcastValidationFixture : public ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>
 {
 public:
@@ -241,6 +256,21 @@ public:
 };
 
 template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ElementwiseMinQuantizedBroadcastValidationFixture : public ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>
+{
+public:
+    template <typename...>
+    void setup(const TensorShape &shape0, const TensorShape &shape1, DataType data_type0, DataType data_type1, DataType output_data_type,
+               QuantizationInfo qinfo0, QuantizationInfo qinfo1, QuantizationInfo qinfo_out)
+
+    {
+        ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(ArithmeticOperation::MIN, shape0, shape1,
+                                                                                             data_type0, data_type1, output_data_type,
+                                                                                             qinfo0, qinfo1, qinfo_out);
+    }
+};
+
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
 class ElementwiseSquaredDiffBroadcastValidationFixture : public ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>
 {
 public:
@@ -280,6 +310,21 @@ public:
                                                                                              qinfo0, qinfo1, qinfo_out);
     }
 };
+
+template <typename TensorType, typename AccessorType, typename FunctionType, typename T>
+class ElementwiseSquaredDiffQuantizedBroadcastValidationFixture : public ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>
+{
+public:
+    template <typename...>
+    void setup(const TensorShape &shape0, const TensorShape &shape1, DataType data_type0, DataType data_type1, DataType output_data_type,
+               QuantizationInfo qinfo0, QuantizationInfo qinfo1, QuantizationInfo qinfo_out)
+
+    {
+        ArithmeticOperationsGenericFixture<TensorType, AccessorType, FunctionType, T>::setup(ArithmeticOperation::SQUARED_DIFF, shape0, shape1,
+                                                                                             data_type0, data_type1, output_data_type,
+                                                                                             qinfo0, qinfo1, qinfo_out);
+    }
+};
 } // namespace validation
 } // namespace test
 } // namespace arm_compute
diff --git a/tests/validation/reference/ElementwiseOperations.cpp b/tests/validation/reference/ElementwiseOperations.cpp
index fe0467fe5e..1f0d42b26e 100644
--- a/tests/validation/reference/ElementwiseOperations.cpp
+++ b/tests/validation/reference/ElementwiseOperations.cpp
@@ -158,6 +158,8 @@ SimpleTensor<uint8_t> arithmetic_operation(ArithmeticOperation op, const SimpleT
     }
 }
 
+template SimpleTensor<int32_t> arithmetic_operation(ArithmeticOperation op, const SimpleTensor<int32_t> &src1, const SimpleTensor<int32_t> &src2, SimpleTensor<int32_t> &dst,
+                                                    ConvertPolicy convert_policy);
 template SimpleTensor<int16_t> arithmetic_operation(ArithmeticOperation op, const SimpleTensor<int16_t> &src1, const SimpleTensor<int16_t> &src2, SimpleTensor<int16_t> &dst,
                                                     ConvertPolicy convert_policy);
 template SimpleTensor<int8_t> arithmetic_operation(ArithmeticOperation op, const SimpleTensor<int8_t> &src1, const SimpleTensor<int8_t> &src2, SimpleTensor<int8_t> &dst,
@@ -175,6 +177,8 @@ SimpleTensor<T> arithmetic_operation(ArithmeticOperation op, const SimpleTensor<
     return dst;
 }
 
+template SimpleTensor<int32_t> arithmetic_operation(ArithmeticOperation op, const SimpleTensor<int32_t> &src1, const SimpleTensor<int32_t> &src2, DataType dst_data_type,
+                                                    ConvertPolicy convert_policy);
 template SimpleTensor<int16_t> arithmetic_operation(ArithmeticOperation op, const SimpleTensor<int16_t> &src1, const SimpleTensor<int16_t> &src2, DataType dst_data_type,
                                                     ConvertPolicy convert_policy);
 template SimpleTensor<int8_t> arithmetic_operation(ArithmeticOperation op, const SimpleTensor<int8_t> &src1, const SimpleTensor<int8_t> &src2, DataType dst_data_type, ConvertPolicy convert_policy);