From 9b461486a863229bb71850f403099ec0713aef68 Mon Sep 17 00:00:00 2001
From: Nikhil Raj <nikhil.raj@arm.com>
Date: Wed, 3 Jul 2019 15:58:31 +0100
Subject: IVGCVSW-3294 Add Neon backend support for PReLU

Change-Id: I2f92f7336d3b4188962a9d08b0e7c6372b1c648f
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
---
 src/backends/neon/workloads/NeonPreluWorkload.cpp | 50 +++++++++++++++++++++++
 1 file changed, 50 insertions(+)
 create mode 100644 src/backends/neon/workloads/NeonPreluWorkload.cpp

(limited to 'src/backends/neon/workloads/NeonPreluWorkload.cpp')

diff --git a/src/backends/neon/workloads/NeonPreluWorkload.cpp b/src/backends/neon/workloads/NeonPreluWorkload.cpp
new file mode 100644
index 0000000000..107090e704
--- /dev/null
+++ b/src/backends/neon/workloads/NeonPreluWorkload.cpp
@@ -0,0 +1,50 @@
+//
+// Copyright © 2017 Arm Ltd. All rights reserved.
+// SPDX-License-Identifier: MIT
+//
+
+#include "NeonPreluWorkload.hpp"
+#include "NeonWorkloadUtils.hpp"
+#include <aclCommon/ArmComputeUtils.hpp>
+
+#include <arm_compute/runtime/NEON/functions/NEPReluLayer.h>
+
+namespace armnn
+{
+
+arm_compute::Status NeonPreluWorkloadValidate(const TensorInfo& input,
+                                              const TensorInfo& alpha,
+                                              const TensorInfo& output)
+{
+    const arm_compute::TensorInfo aclInput = armcomputetensorutils::BuildArmComputeTensorInfo(input);
+    const arm_compute::TensorInfo aclAlpha = armcomputetensorutils::BuildArmComputeTensorInfo(alpha);
+    const arm_compute::TensorInfo aclOutput = armcomputetensorutils::BuildArmComputeTensorInfo(output);
+
+    return arm_compute::NEPReluLayer::validate(&aclInput,
+                                               &aclAlpha,
+                                               &aclOutput);
+}
+
+NeonPreluWorkload::NeonPreluWorkload(const PreluQueueDescriptor& descriptor,
+                                     const WorkloadInfo& info)
+        : BaseWorkload<PreluQueueDescriptor>(descriptor, info)
+{
+    m_Data.ValidateInputsOutputs("NeonPreluWorkload", 1, 1);
+
+    arm_compute::ITensor& input = boost::polymorphic_downcast<IAclTensorHandle*>(m_Data.m_Inputs[0])->GetTensor();
+    arm_compute::ITensor& alpha = boost::polymorphic_downcast<IAclTensorHandle*>(m_Data.m_Inputs[1])->GetTensor();
+    arm_compute::ITensor& output = boost::polymorphic_downcast<IAclTensorHandle*>(m_Data.m_Outputs[0])->GetTensor();
+
+    auto layer = std::make_unique<arm_compute::NEPReluLayer>();
+    layer->configure(&input, &alpha, &output);
+
+    m_PreluLayer.reset(layer.release());
+}
+
+void NeonPreluWorkload::Execute() const
+{
+    ARMNN_SCOPED_PROFILING_EVENT_NEON("NeonPreluWorkload_Execute");
+    m_PreluLayer->run();
+}
+
+} //namespace armnn
\ No newline at end of file
-- 
cgit v1.2.1