COMPMID-3377: Async support to NEElementwiseUnaryLayerKernel kernels/functions

Signed-off-by: Michalis Spyrou <michalis.spyrou@arm.com>
Change-Id: I208287b44ece051e95f891d43a691cb0ac6e56c5
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3419
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

1 files changed, 46 insertions, 1 deletions

diff --git a/src/runtime/NEON/functions/NEPReluLayer.cpp b/src/runtime/NEON/functions/NEPReluLayer.cpp
index 02dfc6f137..1dd01fc162 100644
--- a/src/runtime/NEON/functions/NEPReluLayer.cpp
+++ b/src/runtime/NEON/functions/NEPReluLayer.cpp
@@ -29,7 +29,9 @@
 
 namespace arm_compute
 {
-void NEPReluLayer::configure(const ITensor *input, const ITensor *alpha, ITensor *output)
+namespace experimental
+{
+void NEPReluLayer::configure(const ITensorInfo *input, const ITensorInfo *alpha, ITensorInfo *output)
 {
     auto k = arm_compute::support::cpp14::make_unique<NEArithmeticOperationKernel>();
     k->configure(ArithmeticOperation::PRELU, input, alpha, output);
@@ -40,4 +42,47 @@ Status NEPReluLayer::validate(const ITensorInfo *input, const ITensorInfo *alpha
 {
     return NEArithmeticOperationKernel::validate(ArithmeticOperation::PRELU, input, alpha, output);
 }
+
+MemoryRequirements NEPReluLayer::workspace() const
+{
+    return MemoryRequirements{};
+}
+} // nsamespace experimental
+
+struct NEPReluLayer::Impl
+{
+    const ITensor                              *src_0{ nullptr };
+    const ITensor                              *src_1{ nullptr };
+    ITensor                                    *dst{ nullptr };
+    std::unique_ptr<experimental::NEPReluLayer> op{ nullptr };
+};
+
+NEPReluLayer::NEPReluLayer()
+    : _impl(support::cpp14::make_unique<Impl>())
+{
+}
+NEPReluLayer::NEPReluLayer(NEPReluLayer &&) = default;
+NEPReluLayer &NEPReluLayer::operator=(NEPReluLayer &&) = default;
+NEPReluLayer::~NEPReluLayer()                          = default;
+
+void NEPReluLayer::configure(const ITensor *input, const ITensor *alpha, ITensor *output)
+{
+    _impl->src_0 = input;
+    _impl->src_1 = alpha;
+    _impl->dst   = output;
+    _impl->op    = arm_compute::support::cpp14::make_unique<experimental::NEPReluLayer>();
+    _impl->op->configure(input->info(), alpha->info(), output->info());
+}
+
+void NEPReluLayer::run()
+{
+    const InputTensorMap  src{ { TensorType::ACL_SRC_0, _impl->src_0 }, { TensorType::ACL_SRC_1, _impl->src_1 } };
+    const OutputTensorMap dst{ { TensorType::ACL_DST, _impl->dst } };
+    _impl->op->run(src, dst, {});
+}
+
+Status NEPReluLayer::validate(const ITensorInfo *input, const ITensorInfo *alpha, const ITensorInfo *output)
+{
+    return experimental::NEPReluLayer::validate(input, alpha, output);
+}
 } // namespace arm_compute