COMPMID-2159 Implement Round for CL

Change-Id: I9d3ed4bb3cf7bef39bd0e71b3c6f266ff8187290
Signed-off-by: Usama Arif <usama.arif@arm.com>
Reviewed-on: https://review.mlplatform.org/c/1226
Reviewed-by: Pablo Marquez <pablo.tello@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Tested-by: Arm Jenkins <bsgcomp@arm.com>

6 files changed, 128 insertions, 0 deletions

diff --git a/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h b/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h
index de4bc59240..79e409c03b 100644
--- a/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h
+++ b/arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h
@@ -149,5 +149,25 @@ public:
      */
     static Status validate(const ITensorInfo *input, const ITensorInfo *output);
 };
+
+/** Basic function to get the round (to the nearest even) value of an input tensor. */
+class CLRoundLayer : public ICLSimpleFunction
+{
+public:
+    /** Initialize the function
+     *
+     * @param[in]  input  Input tensor. Data types supported: F16/F32.
+     * @param[out] output Output tensor. Data types supported: same as @p input.
+     */
+    void configure(const ICLTensor *input, ICLTensor *output);
+    /** Static function to check if given info will lead to a valid configuration of @ref CLRoundLayer
+     *
+     * @param[in] input  First tensor input info. Data types supported: F16/F32.
+     * @param[in] output Output tensor info. Data types supported: Same as @p input.
+     *
+     * @return a status
+     */
+    static Status validate(const ITensorInfo *input, const ITensorInfo *output);
+};
 } // namespace arm_compute
 #endif /* __ARM_COMPUTE_CLELEMENTWISEUNARYLAYER_H__ */
diff --git a/src/core/CL/cl_kernels/elementwise_unary.cl b/src/core/CL/cl_kernels/elementwise_unary.cl
index 866b7ee2b0..b496fcf562 100644
--- a/src/core/CL/cl_kernels/elementwise_unary.cl
+++ b/src/core/CL/cl_kernels/elementwise_unary.cl
@@ -38,6 +38,12 @@
 #define fabs_op(input) fabs(input)
 // Calculate natural_log
 #define natural_log_op(input) log(input)
+// Calculate round (Cannot use round function as it rounds halfway cases away from zero).
+#if defined(VEC_SIZE)
+#define round_op(input) CONVERT(CONVERT_SAT_ROUND(input, VEC_DATA_TYPE(int, VEC_SIZE), rte), VEC_DATA_TYPE(DATA_TYPE, VEC_SIZE))
+#else // defined(VEC_SIZE
+#define round_op(input) CONVERT(CONVERT_SAT_ROUND(input, int, rte), DATA_TYPE)
+#endif // defined(VEC_SIZE
 
 /** Applies element wise unary operator in a tensor.
  *
diff --git a/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.cpp b/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.cpp
index 8b8d2965df..03edcef781 100644
--- a/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.cpp
+++ b/src/core/CL/kernels/CLElementWiseUnaryLayerKernel.cpp
@@ -96,6 +96,9 @@ void CLElementWiseUnaryLayerKernel::configure(const ICLTensor *input, ICLTensor
         case ElementWiseUnary::LOG:
             build_opts.add_option("-DOPERATION=natural_log_op");
             break;
+        case ElementWiseUnary::ROUND:
+            build_opts.add_option("-DOPERATION=round_op");
+            break;
         default:
             ARM_COMPUTE_ERROR("Not implemented");
     }
diff --git a/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp b/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp
index b1e18ecd7e..b9ebf6972e 100644
--- a/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp
+++ b/src/runtime/CL/functions/CLElementWiseUnaryLayer.cpp
@@ -95,4 +95,15 @@ Status CLLogLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
     return CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::LOG);
 }
 
+void CLRoundLayer::configure(const ICLTensor *input, ICLTensor *output)
+{
+    auto k = arm_compute::support::cpp14::make_unique<CLElementWiseUnaryLayerKernel>();
+    k->configure(input, output, ElementWiseUnary::ROUND);
+    _kernel = std::move(k);
+}
+Status CLRoundLayer::validate(const ITensorInfo *input, const ITensorInfo *output)
+{
+    return CLElementWiseUnaryLayerKernel::validate(input, output, ElementWiseUnary::ROUND);
+}
+
 } // namespace arm_compute
diff --git a/tests/validation/CL/RoundLayer.cpp b/tests/validation/CL/RoundLayer.cpp
new file mode 100644
index 0000000000..f0dfe8ef23
--- /dev/null
+++ b/tests/validation/CL/RoundLayer.cpp
@@ -0,0 +1,85 @@
+/*
+ * Copyright (c) 2019 ARM Limited.
+ *
+ * SPDX-License-Identifier: MIT
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to
+ * deal in the Software without restriction, including without limitation the
+ * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
+ * sell copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in all
+ * copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ */
+#include "arm_compute/core/Types.h"
+#include "arm_compute/runtime/CL/functions/CLElementWiseUnaryLayer.h"
+#include "arm_compute/runtime/Tensor.h"
+#include "arm_compute/runtime/TensorAllocator.h"
+#include "tests/CL/CLAccessor.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/ElementWiseUnaryFixture.h"
+
+namespace arm_compute
+{
+namespace test
+{
+namespace validation
+{
+TEST_SUITE(CL)
+TEST_SUITE(RoundLayer)
+template <typename T>
+using CLRoundLayerFixture = RoundValidationFixture<CLTensor, CLAccessor, CLRoundLayer, T>;
+
+TEST_SUITE(Float)
+TEST_SUITE(FP16)
+FIXTURE_DATA_TEST_CASE(RunSmall, CLRoundLayerFixture<half>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), framework::dataset::make("DataType",
+                                                                                                       DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, CLRoundLayerFixture<half>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), framework::dataset::make("DataType",
+                                                                                                     DataType::F16)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+TEST_SUITE_END() // FP16
+TEST_SUITE(FP32)
+FIXTURE_DATA_TEST_CASE(RunSmall, CLRoundLayerFixture<float>, framework::DatasetMode::PRECOMMIT, combine(datasets::SmallShapes(), framework::dataset::make("DataType",
+                                                                                                        DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+FIXTURE_DATA_TEST_CASE(RunLarge, CLRoundLayerFixture<float>, framework::DatasetMode::NIGHTLY, combine(datasets::LargeShapes(), framework::dataset::make("DataType",
+                                                                                                      DataType::F32)))
+{
+    // Validate output
+    validate(CLAccessor(_target), _reference);
+}
+
+TEST_SUITE_END() // FP32
+TEST_SUITE_END() // Float
+
+TEST_SUITE_END() // RoundLayer
+TEST_SUITE_END() // CL
+} // namespace validation
+} // namespace test
+} // namespace arm_compute
diff --git a/utils/TypePrinter.h b/utils/TypePrinter.h
index 74dd0bbc35..25c8cd396d 100644
--- a/utils/TypePrinter.h
+++ b/utils/TypePrinter.h
@@ -1531,6 +1531,9 @@ inline ::std::ostream &operator<<(::std::ostream &os, const ElementWiseUnary &op
         case ElementWiseUnary::LOG:
             os << "LOG";
             break;
+        case ElementWiseUnary::ROUND:
+            os << "ROUND";
+            break;
         default:
             ARM_COMPUTE_ERROR("NOT_SUPPORTED!");
     }