From a983e4699082a0b1ef685bab7354f2ad9cd37a44 Mon Sep 17 00:00:00 2001 From: Colm Donelan Date: Wed, 20 May 2020 16:12:19 +0100 Subject: Updating Doxygen documentation for 20.05 release. Change-Id: I4d624343ed5fd6ae269c3d53532903084508fd14 Signed-off-by: Colm Donelan --- 20.05/_loaded_network_8hpp_source.xhtml | 146 ++++++++++++++++++++++++++++++++ 1 file changed, 146 insertions(+) create mode 100644 20.05/_loaded_network_8hpp_source.xhtml (limited to '20.05/_loaded_network_8hpp_source.xhtml') diff --git a/20.05/_loaded_network_8hpp_source.xhtml b/20.05/_loaded_network_8hpp_source.xhtml new file mode 100644 index 0000000000..1446f86d72 --- /dev/null +++ b/20.05/_loaded_network_8hpp_source.xhtml @@ -0,0 +1,146 @@ + + + + + + + + + + + + + +ArmNN: src/armnn/LoadedNetwork.hpp Source File + + + + + + + + + + + + + + + + +
+
+ + + + ArmNN + + + +
+
+  20.05 +
+
+
+ + + + + + + +
+
+ +
+
+
+ +
+ +
+
+ + +
+ +
+ +
+
+
LoadedNetwork.hpp
+
+
+Go to the documentation of this file.
1 //
2 // Copyright © 2017 Arm Ltd. All rights reserved.
3 // SPDX-License-Identifier: MIT
4 //
5 #pragma once
6 
7 #include <armnn/Tensor.hpp>
8 #include <armnn/Types.hpp>
9 
10 #include "Network.hpp"
11 #include "LayerFwd.hpp"
12 #include "Profiling.hpp"
13 
14 #include <armnn/backends/IBackendInternal.hpp>
15 #include <backendsCommon/TensorHandleFactoryRegistry.hpp>
16 #include <backendsCommon/Workload.hpp>
17 #include <backendsCommon/WorkloadFactory.hpp>
18 #include <ProfilingService.hpp>
19 #include <TimelineUtilityMethods.hpp>
20 
21 #include <mutex>
22 #include <unordered_map>
23 
24 namespace cl
25 {
26  class Context;
27  class CommandQueue;
28  class Device;
29 }
30 
31 namespace armnn
32 {
33 
34 class LoadedNetwork
35 {
36 public:
37  using WorkloadQueue = std::vector< std::unique_ptr<IWorkload> >;
38  ~LoadedNetwork(){ FreeWorkingMemory(); }
39 
40  TensorInfo GetInputTensorInfo(LayerBindingId layerId) const;
41  TensorInfo GetOutputTensorInfo(LayerBindingId layerId) const;
42 
43  Status EnqueueWorkload(const InputTensors& inputTensors, const OutputTensors& outputTensors);
44 
45  static std::unique_ptr<LoadedNetwork> MakeLoadedNetwork(std::unique_ptr<OptimizedNetwork> net,
46  std::string & errorMessage,
47  const INetworkProperties& networkProperties,
48  profiling::ProfilingService& profilingService);
49 
50  // NOTE we return by reference as the purpose of this method is only to provide
51  // access to the private m_Profiler and in theory we should not need to increment
52  // the shared_ptr's reference counter
53  const std::shared_ptr<Profiler>& GetProfiler() const { return m_Profiler; }
54 
55  void FreeWorkingMemory();
56 
57  void RegisterDebugCallback(const DebugCallbackFunction& func);
58 
59  void SendNetworkStructure();
60 
61 private:
62  void AllocateWorkingMemory();
63 
64  LoadedNetwork(std::unique_ptr<OptimizedNetwork> net,
65  const INetworkProperties& networkProperties,
66  profiling::ProfilingService& profilingService);
67 
68  void EnqueueInput(const BindableLayer& layer, ITensorHandle* tensorHandle, const TensorInfo& tensorInfo);
69 
70  void EnqueueOutput(const BindableLayer& layer, ITensorHandle* tensorHandle, const TensorInfo& tensorInfo);
71 
72  bool Execute(std::unique_ptr<profiling::TimelineUtilityMethods>& timelineUtils,
73  profiling::ProfilingGuid inferenceGuid);
74 
75 
76  const IWorkloadFactory& GetWorkloadFactory(const Layer& layer) const;
77 
78  using BackendPtrMap = std::unordered_map<BackendId, IBackendInternalUniquePtr>;
79 
80  using WorkloadFactoryWithMemoryManager =
81  std::pair<IBackendInternal::IWorkloadFactoryPtr, IBackendInternal::IMemoryManagerSharedPtr>;
82 
83  using WorkloadFactoryMap = std::unordered_map<BackendId, WorkloadFactoryWithMemoryManager>;
84 
85  BackendPtrMap m_Backends;
86  WorkloadFactoryMap m_WorkloadFactories;
87 
88  std::unique_ptr<OptimizedNetwork> m_OptimizedNetwork;
89  WorkloadQueue m_InputQueue;
90  WorkloadQueue m_WorkloadQueue;
91  WorkloadQueue m_OutputQueue;
92  std::shared_ptr<Profiler> m_Profiler;
93 
94  mutable std::mutex m_WorkingMemMutex;
95 
96  bool m_IsWorkingMemAllocated=false;
97  bool m_IsImportEnabled=false;
98  bool m_IsExportEnabled=false;
99 
100  TensorHandleFactoryRegistry m_TensorHandleFactoryRegistry;
101 
102  profiling::ProfilingService& m_ProfilingService;
103 };
104 
105 }
+
Tensor.hpp
+
armnn::LoadedNetwork::WorkloadQueue
std::vector< std::unique_ptr< IWorkload > > WorkloadQueue
Definition: LoadedNetwork.hpp:37
+
WorkloadFactory.hpp
+ + + + +
armnn::InputTensors
std::vector< std::pair< LayerBindingId, class ConstTensor > > InputTensors
Definition: Tensor.hpp:225
+
armnn::LoadedNetwork::GetProfiler
const std::shared_ptr< Profiler > & GetProfiler() const
Definition: LoadedNetwork.hpp:53
+
armnn
Copyright (c) 2020 ARM Limited.
Definition: 00_introduction.dox:25
+
armnn::DebugCallbackFunction
std::function< void(LayerGuid guid, unsigned int slotIndex, ITensorHandle *tensorHandle)> DebugCallbackFunction
Define the type of callback for the Debug layer to call.
Definition: Types.hpp:244
+
armnn::LayerBindingId
int LayerBindingId
Type of identifiers for bindable layers (inputs, outputs).
Definition: Types.hpp:171
+ + + +
Types.hpp
+ +
armnn::OutputTensors
std::vector< std::pair< LayerBindingId, class Tensor > > OutputTensors
Definition: Tensor.hpp:226
+
armnn::Status
Status
enumeration
Definition: Types.hpp:26
+
profilingService
armnn::profiling::ProfilingService profilingService
Definition: FileOnlyProfilingDecoratorTests.cpp:116
+ + +
Workload.hpp
+ +
TimelineUtilityMethods.hpp
+
LayerFwd.hpp
+
Network.hpp
+
Profiling.hpp
+
TensorHandleFactoryRegistry.hpp
+
armnn::GetInputTensorInfo
TensorInfo GetInputTensorInfo(const Network *network)
Definition: QuantizerTest.cpp:336
+ +
ProfilingService.hpp
+
+
+ + + + -- cgit v1.2.1