diff options
| author | Nikhil Raj <nikhil.raj@arm.com> | 2022-08-19 15:23:36 +0100 |
|---|---|---|
| committer | Nikhil Raj <nikhil.raj@arm.com> | 2022-08-19 15:23:36 +0100 |
| commit | 7bfd38a721360183f3392f9ab35db18a0dd7fef8 (patch) | |
| tree | 5b4da2f2e88636c939afbafa2571170297114e40 /22.08/_cl_import_tensor_handle_tests_8cpp.xhtml | |
| parent | d5d43d82c0137e08553e44345c609cdd1a7931c7 (diff) | |
| download | armnn-7bfd38a721360183f3392f9ab35db18a0dd7fef8.tar.gz | |
Update Doxygen for 22.08 Release
Signed-off-by: Nikhil Raj <nikhil.raj@arm.com>
Change-Id: I4789fe868e0492839be1482e5cee3642ed90d756
Diffstat (limited to '22.08/_cl_import_tensor_handle_tests_8cpp.xhtml')
| -rw-r--r-- | 22.08/_cl_import_tensor_handle_tests_8cpp.xhtml | 213 |
1 files changed, 213 insertions, 0 deletions
diff --git a/22.08/_cl_import_tensor_handle_tests_8cpp.xhtml b/22.08/_cl_import_tensor_handle_tests_8cpp.xhtml new file mode 100644 index 0000000000..5c37d03428 --- /dev/null +++ b/22.08/_cl_import_tensor_handle_tests_8cpp.xhtml @@ -0,0 +1,213 @@ +<!-- Copyright (c) 2020 ARM Limited. --> +<!-- --> +<!-- SPDX-License-Identifier: MIT --> +<!-- --> +<!-- HTML header for doxygen 1.8.13--> +<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> +<html xmlns="http://www.w3.org/1999/xhtml"> +<head> +<meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/> +<meta http-equiv="X-UA-Compatible" content="IE=9"/> +<meta name="generator" content="Doxygen 1.8.13"/> +<meta name="robots" content="NOINDEX, NOFOLLOW" /> +<meta name="viewport" content="width=device-width, initial-scale=1"/> +<title>ArmNN: src/backends/cl/test/ClImportTensorHandleTests.cpp File Reference</title> +<link href="tabs.css" rel="stylesheet" type="text/css"/> +<script type="text/javascript" src="jquery.js"></script> +<script type="text/javascript" src="dynsections.js"></script> +<link href="navtree.css" rel="stylesheet" type="text/css"/> +<script type="text/javascript" src="resize.js"></script> +<script type="text/javascript" src="navtreedata.js"></script> +<script type="text/javascript" src="navtree.js"></script> +<script type="text/javascript"> + $(document).ready(initResizable); +</script> +<link href="search/search.css" rel="stylesheet" type="text/css"/> +<script type="text/javascript" src="search/searchdata.js"></script> +<script type="text/javascript" src="search/search.js"></script> +<script type="text/x-mathjax-config"> + MathJax.Hub.Config({ + extensions: ["tex2jax.js"], + jax: ["input/TeX","output/HTML-CSS"], +}); +</script><script type="text/javascript" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script> +<link href="doxygen.css" rel="stylesheet" type="text/css" /> +<link href="stylesheet.css" rel="stylesheet" type="text/css"/> +</head> +<body> +<div id="top"><!-- do not remove this div, it is closed by doxygen! --> +<div id="titlearea"> +<table cellspacing="0" cellpadding="0"> + <tbody> + <tr style="height: 56px;"> + <img alt="ArmNN" src="Arm_NN_horizontal_blue.png" style="max-width: 10rem; margin-top: .5rem; margin-left 10px"/> + <td style="padding-left: 0.5em;"> + <div id="projectname"> +  <span id="projectnumber">22.08</span> + </div> + </td> + </tr> + </tbody> +</table> +</div> +<!-- end header part --> +<!-- Generated by Doxygen 1.8.13 --> +<script type="text/javascript"> +var searchBox = new SearchBox("searchBox", "search",false,'Search'); +</script> +<script type="text/javascript" src="menudata.js"></script> +<script type="text/javascript" src="menu.js"></script> +<script type="text/javascript"> +$(function() { + initMenu('',true,false,'search.php','Search'); + $(document).ready(function() { init_search(); }); +}); +</script> +<div id="main-nav"></div> +</div><!-- top --> +<div id="side-nav" class="ui-resizable side-nav-resizable"> + <div id="nav-tree"> + <div id="nav-tree-contents"> + <div id="nav-sync" class="sync"></div> + </div> + </div> + <div id="splitbar" style="-moz-user-select:none;" + class="ui-resizable-handle"> + </div> +</div> +<script type="text/javascript"> +$(document).ready(function(){initNavTree('_cl_import_tensor_handle_tests_8cpp.xhtml','');}); +</script> +<div id="doc-content"> +<!-- window showing the filter options --> +<div id="MSearchSelectWindow" + onmouseover="return searchBox.OnSearchSelectShow()" + onmouseout="return searchBox.OnSearchSelectHide()" + onkeydown="return searchBox.OnSearchSelectKey(event)"> +</div> + +<!-- iframe showing the search results (closed by default) --> +<div id="MSearchResultsWindow"> +<iframe src="javascript:void(0)" frameborder="0" + name="MSearchResults" id="MSearchResults"> +</iframe> +</div> + +<div class="header"> + <div class="summary"> +<a href="#func-members">Functions</a> </div> + <div class="headertitle"> +<div class="title">ClImportTensorHandleTests.cpp File Reference</div> </div> +</div><!--header--> +<div class="contents"> +<div class="textblock"><code>#include <arm_compute/runtime/CL/functions/CLActivationLayer.h></code><br /> +<code>#include <<a class="el" href="_cl_import_tensor_handle_8hpp_source.xhtml">cl/ClImportTensorHandle.hpp</a>></code><br /> +<code>#include <<a class="el" href="_cl_import_tensor_handle_factory_8hpp_source.xhtml">cl/ClImportTensorHandleFactory.hpp</a>></code><br /> +<code>#include <<a class="el" href="_cl_context_control_fixture_8hpp_source.xhtml">cl/test/ClContextControlFixture.hpp</a>></code><br /> +<code>#include <doctest/doctest.h></code><br /> +<code>#include <<a class="el" href="_i_runtime_8hpp_source.xhtml">armnn/IRuntime.hpp</a>></code><br /> +<code>#include <<a class="el" href="_i_network_8hpp_source.xhtml">armnn/INetwork.hpp</a>></code><br /> +<code>#include "<a class="el" href="_network_8hpp_source.xhtml">Network.hpp</a>"</code><br /> +</div> +<p><a href="_cl_import_tensor_handle_tests_8cpp_source.xhtml">Go to the source code of this file.</a></p> +<table class="memberdecls"> +<tr class="heading"><td colspan="2"><h2 class="groupheader"><a name="func-members"></a> +Functions</h2></td></tr> +<tr class="memitem:a98b876489de8b7d460ee756beac83891"><td class="memItemLeft" align="right" valign="top"> </td><td class="memItemRight" valign="bottom"><a class="el" href="_cl_import_tensor_handle_tests_8cpp.xhtml#a98b876489de8b7d460ee756beac83891">TEST_SUITE</a> ("ClImportTensorHandleTests")</td></tr> +<tr class="separator:a98b876489de8b7d460ee756beac83891"><td class="memSeparator" colspan="2"> </td></tr> +</table> +<h2 class="groupheader">Function Documentation</h2> +<a id="a98b876489de8b7d460ee756beac83891"></a> +<h2 class="memtitle"><span class="permalink"><a href="#a98b876489de8b7d460ee756beac83891">◆ </a></span>TEST_SUITE()</h2> + +<div class="memitem"> +<div class="memproto"> + <table class="memname"> + <tr> + <td class="memname">TEST_SUITE </td> + <td>(</td> + <td class="paramtype">"ClImportTensorHandleTests" </td> + <td class="paramname"></td><td>)</td> + <td></td> + </tr> + </table> +</div><div class="memdoc"> + +<p class="definition">Definition at line <a class="el" href="_cl_import_tensor_handle_tests_8cpp_source.xhtml#l00020">20</a> of file <a class="el" href="_cl_import_tensor_handle_tests_8cpp_source.xhtml">ClImportTensorHandleTests.cpp</a>.</p> + +<p class="reference">References <a class="el" href="_network_8cpp_source.xhtml#l02079">NetworkImpl::AddConvertFp16ToFp32Layer()</a>, <a class="el" href="_network_8cpp_source.xhtml#l02084">NetworkImpl::AddConvertFp32ToFp16Layer()</a>, <a class="el" href="_network_8cpp_source.xhtml#l01920">NetworkImpl::AddInputLayer()</a>, <a class="el" href="_network_8cpp_source.xhtml#l02224">NetworkImpl::AddOutputLayer()</a>, <a class="el" href="_profiling_8cpp_source.xhtml#l00604">IProfiler::AnalyzeEventsAndWriteResults()</a>, <a class="el" href="_assert_8hpp_source.xhtml#l00014">ARMNN_ASSERT</a>, <a class="el" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">IOutputSlot::Connect()</a>, <a class="el" href="_runtime_8cpp_source.xhtml#l00049">IRuntime::Create()</a>, <a class="el" href="_network_8cpp_source.xhtml#l00475">INetwork::Create()</a>, <a class="el" href="_cl_import_tensor_handle_factory_8cpp_source.xhtml#l00056">ClImportTensorHandleFactory::CreateTensorHandle()</a>, <a class="el" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a26e6ed77470c6f2f830ecf874e6c0d55">armnn::Float16</a>, <a class="el" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a166495adc0d0f53bee6baecc577f5204">armnn::Float32</a>, <a class="el" href="_network_8hpp_source.xhtml#l00037">NetworkImpl::GetGraph()</a>, <a class="el" href="_tensor_8hpp_source.xhtml#l00295">BaseTensor< MemoryType >::GetInfo()</a>, <a class="el" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">IConnectableLayer::GetInputSlot()</a>, <a class="el" href="_profiling_8cpp_source.xhtml#l00572">ProfilerManager::GetInstance()</a>, <a class="el" href="_tensor_8hpp_source.xhtml#l00196">TensorInfo::GetNumElements()</a>, <a class="el" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">IConnectableLayer::GetOutputSlot()</a>, <a class="el" href="_profiling_8cpp_source.xhtml#l00584">ProfilerManager::GetProfiler()</a>, <a class="el" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::GpuAcc</a>, <a class="el" href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">armnn::info</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00545">Convolution2dDescriptor::m_DataLayout</a>, <a class="el" href="_i_network_8hpp_source.xhtml#l00233">OptimizerOptions::m_ExportEnabled</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00059">ActivationDescriptor::m_Function</a>, <a class="el" href="_i_network_8hpp_source.xhtml#l00224">OptimizerOptions::m_ImportEnabled</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00533">Convolution2dDescriptor::m_PadBottom</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00527">Convolution2dDescriptor::m_PadLeft</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00529">Convolution2dDescriptor::m_PadRight</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00531">Convolution2dDescriptor::m_PadTop</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00535">Convolution2dDescriptor::m_StrideX</a>, <a class="el" href="_descriptors_8hpp_source.xhtml#l00537">Convolution2dDescriptor::m_StrideY</a>, <a class="el" href="namespacearmnn.xhtml#a14fcd7f88d11cea0a018269dca5f9277a1131a914388fac73e5f07b0ba0aad523">armnn::Malloc</a>, <a class="el" href="namespacearmnn.xhtml#ad1d5cce2d9e9a5d61c243e5c989112e0ad066db54b89b0912e7e7c6da51e2da51">armnn::NHWC</a>, <a class="el" href="_network_8cpp_source.xhtml#l01864">armnn::Optimize()</a>, <a class="el" href="_profiling_8cpp_source.xhtml#l00609">IProfiler::Print()</a>, <a class="el" href="namespacearmnn.xhtml#a56297e0f7b215eea46c818cb7528d9eaa3d90c0a5ab3fcf8e6f6608cb3d3a1559">armnn::ReLu</a>, <a class="el" href="_tensor_8cpp_source.xhtml#l00514">TensorInfo::SetConstant()</a>, <a class="el" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">IOutputSlot::SetTensorInfo()</a>, <a class="el" href="_mem_copy_tests_8cpp_source.xhtml#l00089">TEST_CASE_FIXTURE()</a>, and <a class="el" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeaec0fc0100c4fc1ce4eea230c3dc10360">armnn::Undefined</a>.</p> +<div class="fragment"><div class="line"><a name="l00021"></a><span class="lineno"> 21</span> {</div><div class="line"><a name="l00022"></a><span class="lineno"> 22</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClMallocImport"</span>)</div><div class="line"><a name="l00023"></a><span class="lineno"> 23</span> {</div><div class="line"><a name="l00024"></a><span class="lineno"> 24</span>  <a class="code" href="classarmnn_1_1_cl_import_tensor_handle_factory.xhtml">ClImportTensorHandleFactory</a> handleFactory(static_cast<MemorySourceFlags>(MemorySource::Malloc),</div><div class="line"><a name="l00025"></a><span class="lineno"> 25</span>  static_cast<MemorySourceFlags>(MemorySource::Malloc));</div><div class="line"><a name="l00026"></a><span class="lineno"> 26</span> </div><div class="line"><a name="l00027"></a><span class="lineno"> 27</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> <a class="code" href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>({ 1, 24, 16, 3 }, DataType::Float32);</div><div class="line"><a name="l00028"></a><span class="lineno"> 28</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = <a class="code" href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>.GetNumElements();</div><div class="line"><a name="l00029"></a><span class="lineno"> 29</span> </div><div class="line"><a name="l00030"></a><span class="lineno"> 30</span>  <span class="comment">// create TensorHandle for memory import</span></div><div class="line"><a name="l00031"></a><span class="lineno"> 31</span>  <span class="keyword">auto</span> handle = handleFactory.CreateTensorHandle(info);</div><div class="line"><a name="l00032"></a><span class="lineno"> 32</span> </div><div class="line"><a name="l00033"></a><span class="lineno"> 33</span>  <span class="comment">// Get CLtensor</span></div><div class="line"><a name="l00034"></a><span class="lineno"> 34</span>  arm_compute::CLTensor& tensor = PolymorphicDowncast<ClImportTensorHandle*>(handle.get())->GetTensor();</div><div class="line"><a name="l00035"></a><span class="lineno"> 35</span> </div><div class="line"><a name="l00036"></a><span class="lineno"> 36</span>  <span class="comment">// Create and configure activation function</span></div><div class="line"><a name="l00037"></a><span class="lineno"> 37</span>  <span class="keyword">const</span> arm_compute::ActivationLayerInfo act_info(arm_compute::ActivationLayerInfo::ActivationFunction::RELU);</div><div class="line"><a name="l00038"></a><span class="lineno"> 38</span>  arm_compute::CLActivationLayer act_func;</div><div class="line"><a name="l00039"></a><span class="lineno"> 39</span>  act_func.configure(&tensor, <span class="keyword">nullptr</span>, act_info);</div><div class="line"><a name="l00040"></a><span class="lineno"> 40</span> </div><div class="line"><a name="l00041"></a><span class="lineno"> 41</span>  <span class="comment">// Allocate user memory</span></div><div class="line"><a name="l00042"></a><span class="lineno"> 42</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> totalBytes = tensor.info()->total_size();</div><div class="line"><a name="l00043"></a><span class="lineno"> 43</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00044"></a><span class="lineno"> 44</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00045"></a><span class="lineno"> 45</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l00046"></a><span class="lineno"> 46</span>  <span class="keyword">auto</span> testData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00047"></a><span class="lineno"> 47</span>  <span class="keywordtype">void</span>* alignedPtr = testData.get();</div><div class="line"><a name="l00048"></a><span class="lineno"> 48</span>  CHECK(std::align(alignment, totalBytes, alignedPtr, space));</div><div class="line"><a name="l00049"></a><span class="lineno"> 49</span> </div><div class="line"><a name="l00050"></a><span class="lineno"> 50</span>  <span class="comment">// Import memory</span></div><div class="line"><a name="l00051"></a><span class="lineno"> 51</span>  CHECK(handle->Import(alignedPtr, <a class="code" href="namespacearmnn.xhtml#a14fcd7f88d11cea0a018269dca5f9277a1131a914388fac73e5f07b0ba0aad523">armnn::MemorySource::Malloc</a>));</div><div class="line"><a name="l00052"></a><span class="lineno"> 52</span> </div><div class="line"><a name="l00053"></a><span class="lineno"> 53</span>  <span class="comment">// Input with negative values</span></div><div class="line"><a name="l00054"></a><span class="lineno"> 54</span>  <span class="keyword">auto</span>* typedPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedPtr);</div><div class="line"><a name="l00055"></a><span class="lineno"> 55</span>  std::fill_n(typedPtr, numElements, -5.0f);</div><div class="line"><a name="l00056"></a><span class="lineno"> 56</span> </div><div class="line"><a name="l00057"></a><span class="lineno"> 57</span>  <span class="comment">// Execute function and sync</span></div><div class="line"><a name="l00058"></a><span class="lineno"> 58</span>  act_func.run();</div><div class="line"><a name="l00059"></a><span class="lineno"> 59</span>  arm_compute::CLScheduler::get().sync();</div><div class="line"><a name="l00060"></a><span class="lineno"> 60</span> </div><div class="line"><a name="l00061"></a><span class="lineno"> 61</span>  <span class="comment">// Validate result by checking that the output has no negative values</span></div><div class="line"><a name="l00062"></a><span class="lineno"> 62</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> i = 0; i < numElements; ++i)</div><div class="line"><a name="l00063"></a><span class="lineno"> 63</span>  {</div><div class="line"><a name="l00064"></a><span class="lineno"> 64</span>  CHECK(typedPtr[i] == 0);</div><div class="line"><a name="l00065"></a><span class="lineno"> 65</span>  }</div><div class="line"><a name="l00066"></a><span class="lineno"> 66</span> }</div><div class="line"><a name="l00067"></a><span class="lineno"> 67</span> </div><div class="line"><a name="l00068"></a><span class="lineno"> 68</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClIncorrectMemorySourceImport"</span>)</div><div class="line"><a name="l00069"></a><span class="lineno"> 69</span> {</div><div class="line"><a name="l00070"></a><span class="lineno"> 70</span>  <a class="code" href="classarmnn_1_1_cl_import_tensor_handle_factory.xhtml">ClImportTensorHandleFactory</a> handleFactory(static_cast<MemorySourceFlags>(MemorySource::Malloc),</div><div class="line"><a name="l00071"></a><span class="lineno"> 71</span>  static_cast<MemorySourceFlags>(MemorySource::Malloc));</div><div class="line"><a name="l00072"></a><span class="lineno"> 72</span> </div><div class="line"><a name="l00073"></a><span class="lineno"> 73</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> <a class="code" href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>({ 1, 24, 16, 3 }, DataType::Float32);</div><div class="line"><a name="l00074"></a><span class="lineno"> 74</span> </div><div class="line"><a name="l00075"></a><span class="lineno"> 75</span>  <span class="comment">// create TensorHandle for memory import</span></div><div class="line"><a name="l00076"></a><span class="lineno"> 76</span>  <span class="keyword">auto</span> handle = handleFactory.CreateTensorHandle(info);</div><div class="line"><a name="l00077"></a><span class="lineno"> 77</span> </div><div class="line"><a name="l00078"></a><span class="lineno"> 78</span>  <span class="comment">// Get CLtensor</span></div><div class="line"><a name="l00079"></a><span class="lineno"> 79</span>  arm_compute::CLTensor& tensor = PolymorphicDowncast<ClImportTensorHandle*>(handle.get())->GetTensor();</div><div class="line"><a name="l00080"></a><span class="lineno"> 80</span> </div><div class="line"><a name="l00081"></a><span class="lineno"> 81</span>  <span class="comment">// Allocate user memory</span></div><div class="line"><a name="l00082"></a><span class="lineno"> 82</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> totalBytes = tensor.info()->total_size();</div><div class="line"><a name="l00083"></a><span class="lineno"> 83</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00084"></a><span class="lineno"> 84</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00085"></a><span class="lineno"> 85</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l00086"></a><span class="lineno"> 86</span>  <span class="keyword">auto</span> testData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  <span class="keywordtype">void</span>* alignedPtr = testData.get();</div><div class="line"><a name="l00088"></a><span class="lineno"> 88</span>  CHECK(std::align(alignment, totalBytes, alignedPtr, space));</div><div class="line"><a name="l00089"></a><span class="lineno"> 89</span> </div><div class="line"><a name="l00090"></a><span class="lineno"> 90</span>  <span class="comment">// Import memory</span></div><div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  CHECK_THROWS_AS(handle->Import(alignedPtr, <a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeaec0fc0100c4fc1ce4eea230c3dc10360">armnn::MemorySource::Undefined</a>), <a class="code" href="classarmnn_1_1_memory_import_exception.xhtml">MemoryImportException</a>);</div><div class="line"><a name="l00092"></a><span class="lineno"> 92</span> }</div><div class="line"><a name="l00093"></a><span class="lineno"> 93</span> </div><div class="line"><a name="l00094"></a><span class="lineno"> 94</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClInvalidMemorySourceImport"</span>)</div><div class="line"><a name="l00095"></a><span class="lineno"> 95</span> {</div><div class="line"><a name="l00096"></a><span class="lineno"> 96</span>  <a class="code" href="namespacearmnn.xhtml#a14fcd7f88d11cea0a018269dca5f9277">MemorySource</a> invalidMemSource = <span class="keyword">static_cast<</span><a class="code" href="namespacearmnn.xhtml#a14fcd7f88d11cea0a018269dca5f9277">MemorySource</a><span class="keyword">></span>(256);</div><div class="line"><a name="l00097"></a><span class="lineno"> 97</span>  <a class="code" href="classarmnn_1_1_cl_import_tensor_handle_factory.xhtml">ClImportTensorHandleFactory</a> handleFactory(static_cast<MemorySourceFlags>(invalidMemSource),</div><div class="line"><a name="l00098"></a><span class="lineno"> 98</span>  static_cast<MemorySourceFlags>(invalidMemSource));</div><div class="line"><a name="l00099"></a><span class="lineno"> 99</span> </div><div class="line"><a name="l00100"></a><span class="lineno"> 100</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> <a class="code" href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>({ 1, 2, 2, 1 }, DataType::Float32);</div><div class="line"><a name="l00101"></a><span class="lineno"> 101</span> </div><div class="line"><a name="l00102"></a><span class="lineno"> 102</span>  <span class="comment">// create TensorHandle for memory import</span></div><div class="line"><a name="l00103"></a><span class="lineno"> 103</span>  <span class="keyword">auto</span> handle = handleFactory.CreateTensorHandle(info);</div><div class="line"><a name="l00104"></a><span class="lineno"> 104</span> </div><div class="line"><a name="l00105"></a><span class="lineno"> 105</span>  <span class="comment">// Allocate user memory</span></div><div class="line"><a name="l00106"></a><span class="lineno"> 106</span>  std::vector<float> inputData</div><div class="line"><a name="l00107"></a><span class="lineno"> 107</span>  {</div><div class="line"><a name="l00108"></a><span class="lineno"> 108</span>  1.0f, 2.0f, 3.0f, 4.0f</div><div class="line"><a name="l00109"></a><span class="lineno"> 109</span>  };</div><div class="line"><a name="l00110"></a><span class="lineno"> 110</span> </div><div class="line"><a name="l00111"></a><span class="lineno"> 111</span>  <span class="comment">// Import non-support memory</span></div><div class="line"><a name="l00112"></a><span class="lineno"> 112</span>  CHECK_THROWS_AS(handle->Import(inputData.data(), invalidMemSource), <a class="code" href="classarmnn_1_1_memory_import_exception.xhtml">MemoryImportException</a>);</div><div class="line"><a name="l00113"></a><span class="lineno"> 113</span> }</div><div class="line"><a name="l00114"></a><span class="lineno"> 114</span> </div><div class="line"><a name="l00115"></a><span class="lineno"> 115</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClImportEndToEnd"</span>)</div><div class="line"><a name="l00116"></a><span class="lineno"> 116</span> {</div><div class="line"><a name="l00117"></a><span class="lineno"> 117</span>  <span class="comment">// Create runtime in which test will run</span></div><div class="line"><a name="l00118"></a><span class="lineno"> 118</span>  <a class="code" href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">IRuntime::CreationOptions</a> options;</div><div class="line"><a name="l00119"></a><span class="lineno"> 119</span>  <a class="code" href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">IRuntimePtr</a> runtime(<a class="code" href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a>(options));</div><div class="line"><a name="l00120"></a><span class="lineno"> 120</span> </div><div class="line"><a name="l00121"></a><span class="lineno"> 121</span>  <span class="comment">// build up the structure of the network</span></div><div class="line"><a name="l00122"></a><span class="lineno"> 122</span>  <a class="code" href="namespacearmnn.xhtml#ace74f6f9feb95a964a49d79458232703">INetworkPtr</a> net(INetwork::Create());</div><div class="line"><a name="l00123"></a><span class="lineno"> 123</span> </div><div class="line"><a name="l00124"></a><span class="lineno"> 124</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* input = net->AddInputLayer(0, <span class="stringliteral">"Input"</span>);</div><div class="line"><a name="l00125"></a><span class="lineno"> 125</span> </div><div class="line"><a name="l00126"></a><span class="lineno"> 126</span>  <a class="code" href="structarmnn_1_1_activation_descriptor.xhtml">ActivationDescriptor</a> descriptor;</div><div class="line"><a name="l00127"></a><span class="lineno"> 127</span>  descriptor.<a class="code" href="structarmnn_1_1_activation_descriptor.xhtml#af10fa7883e3579950f477bee92a64844">m_Function</a> = ActivationFunction::ReLu;</div><div class="line"><a name="l00128"></a><span class="lineno"> 128</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* activation = net->AddActivationLayer(descriptor, <span class="stringliteral">"Activation"</span>);</div><div class="line"><a name="l00129"></a><span class="lineno"> 129</span> </div><div class="line"><a name="l00130"></a><span class="lineno"> 130</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* output = net->AddOutputLayer(0, <span class="stringliteral">"Output"</span>);</div><div class="line"><a name="l00131"></a><span class="lineno"> 131</span> </div><div class="line"><a name="l00132"></a><span class="lineno"> 132</span>  input-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(activation-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00133"></a><span class="lineno"> 133</span>  activation-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(output-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00134"></a><span class="lineno"> 134</span> </div><div class="line"><a name="l00135"></a><span class="lineno"> 135</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> tensorInfo = <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a>({ 1, 24, 16, 3 }, DataType::Float32);</div><div class="line"><a name="l00136"></a><span class="lineno"> 136</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = tensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8846406ac37fbd2204f0be16ee05d5b7">GetNumElements</a>();</div><div class="line"><a name="l00137"></a><span class="lineno"> 137</span>  <span class="keywordtype">size_t</span> totalBytes = numElements * <span class="keyword">sizeof</span>(float);</div><div class="line"><a name="l00138"></a><span class="lineno"> 138</span> </div><div class="line"><a name="l00139"></a><span class="lineno"> 139</span>  input-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(tensorInfo);</div><div class="line"><a name="l00140"></a><span class="lineno"> 140</span>  activation-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(tensorInfo);</div><div class="line"><a name="l00141"></a><span class="lineno"> 141</span> </div><div class="line"><a name="l00142"></a><span class="lineno"> 142</span>  <span class="comment">// Optimize the network</span></div><div class="line"><a name="l00143"></a><span class="lineno"> 143</span>  <a class="code" href="structarmnn_1_1_optimizer_options.xhtml">OptimizerOptions</a> optOptions;</div><div class="line"><a name="l00144"></a><span class="lineno"> 144</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">m_ImportEnabled</a> = <span class="keyword">true</span>;</div><div class="line"><a name="l00145"></a><span class="lineno"> 145</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">m_ExportEnabled</a> = <span class="keyword">true</span>;</div><div class="line"><a name="l00146"></a><span class="lineno"> 146</span>  std::vector<armnn::BackendId> backends = {<a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a>};</div><div class="line"><a name="l00147"></a><span class="lineno"> 147</span>  <a class="code" href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">IOptimizedNetworkPtr</a> optNet = <a class="code" href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">Optimize</a>(*net, backends, runtime->GetDeviceSpec(), optOptions);</div><div class="line"><a name="l00148"></a><span class="lineno"> 148</span>  CHECK(optNet);</div><div class="line"><a name="l00149"></a><span class="lineno"> 149</span> </div><div class="line"><a name="l00150"></a><span class="lineno"> 150</span>  <span class="comment">// Loads it into the runtime.</span></div><div class="line"><a name="l00151"></a><span class="lineno"> 151</span>  <a class="code" href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">NetworkId</a> netId;</div><div class="line"><a name="l00152"></a><span class="lineno"> 152</span>  std::string ignoredErrorMessage;</div><div class="line"><a name="l00153"></a><span class="lineno"> 153</span>  <span class="comment">// Enable Importing</span></div><div class="line"><a name="l00154"></a><span class="lineno"> 154</span>  <a class="code" href="structarmnn_1_1_i_network_properties.xhtml">INetworkProperties</a> networkProperties(<span class="keyword">false</span>, MemorySource::Malloc, MemorySource::Malloc);</div><div class="line"><a name="l00155"></a><span class="lineno"> 155</span>  runtime->LoadNetwork(netId, std::move(optNet), ignoredErrorMessage, networkProperties);</div><div class="line"><a name="l00156"></a><span class="lineno"> 156</span> </div><div class="line"><a name="l00157"></a><span class="lineno"> 157</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l00158"></a><span class="lineno"> 158</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00159"></a><span class="lineno"> 159</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00160"></a><span class="lineno"> 160</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l00161"></a><span class="lineno"> 161</span>  <span class="keyword">auto</span> inputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00162"></a><span class="lineno"> 162</span>  <span class="keywordtype">void</span>* alignedInputPtr = inputData.get();</div><div class="line"><a name="l00163"></a><span class="lineno"> 163</span>  CHECK(std::align(alignment, totalBytes, alignedInputPtr, space));</div><div class="line"><a name="l00164"></a><span class="lineno"> 164</span> </div><div class="line"><a name="l00165"></a><span class="lineno"> 165</span>  <span class="comment">// Input with negative values</span></div><div class="line"><a name="l00166"></a><span class="lineno"> 166</span>  <span class="keyword">auto</span>* intputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedInputPtr);</div><div class="line"><a name="l00167"></a><span class="lineno"> 167</span>  std::fill_n(intputPtr, numElements, -5.0f);</div><div class="line"><a name="l00168"></a><span class="lineno"> 168</span> </div><div class="line"><a name="l00169"></a><span class="lineno"> 169</span>  <span class="keyword">auto</span> outputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00170"></a><span class="lineno"> 170</span>  <span class="keywordtype">void</span>* alignedOutputPtr = outputData.get();</div><div class="line"><a name="l00171"></a><span class="lineno"> 171</span>  CHECK(std::align(alignment, totalBytes, alignedOutputPtr, space));</div><div class="line"><a name="l00172"></a><span class="lineno"> 172</span>  <span class="keyword">auto</span>* outputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00173"></a><span class="lineno"> 173</span>  std::fill_n(outputPtr, numElements, -10.0f);</div><div class="line"><a name="l00174"></a><span class="lineno"> 174</span> </div><div class="line"><a name="l00175"></a><span class="lineno"> 175</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> inputTensorInfo = runtime->GetInputTensorInfo(netId, 0);</div><div class="line"><a name="l00176"></a><span class="lineno"> 176</span>  inputTensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00177"></a><span class="lineno"> 177</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensors</div><div class="line"><a name="l00178"></a><span class="lineno"> 178</span>  {</div><div class="line"><a name="l00179"></a><span class="lineno"> 179</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, alignedInputPtr)},</div><div class="line"><a name="l00180"></a><span class="lineno"> 180</span>  };</div><div class="line"><a name="l00181"></a><span class="lineno"> 181</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensors</div><div class="line"><a name="l00182"></a><span class="lineno"> 182</span>  {</div><div class="line"><a name="l00183"></a><span class="lineno"> 183</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), alignedOutputPtr)}</div><div class="line"><a name="l00184"></a><span class="lineno"> 184</span>  };</div><div class="line"><a name="l00185"></a><span class="lineno"> 185</span> </div><div class="line"><a name="l00186"></a><span class="lineno"> 186</span>  runtime->GetProfiler(netId)->EnableProfiling(<span class="keyword">true</span>);</div><div class="line"><a name="l00187"></a><span class="lineno"> 187</span> </div><div class="line"><a name="l00188"></a><span class="lineno"> 188</span>  <span class="comment">// Do the inference</span></div><div class="line"><a name="l00189"></a><span class="lineno"> 189</span>  runtime->EnqueueWorkload(netId, inputTensors, outputTensors);</div><div class="line"><a name="l00190"></a><span class="lineno"> 190</span> </div><div class="line"><a name="l00191"></a><span class="lineno"> 191</span>  <span class="comment">// Retrieve the Profiler.Print() output to get the workload execution</span></div><div class="line"><a name="l00192"></a><span class="lineno"> 192</span>  <a class="code" href="classarmnn_1_1_profiler_manager.xhtml">ProfilerManager</a>& profilerManager = <a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a>();</div><div class="line"><a name="l00193"></a><span class="lineno"> 193</span>  std::stringstream ss;</div><div class="line"><a name="l00194"></a><span class="lineno"> 194</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#a038bb767bbc6abc0ee0d9b509616b896">Print</a>(ss);;</div><div class="line"><a name="l00195"></a><span class="lineno"> 195</span>  std::string dump = ss.str();</div><div class="line"><a name="l00196"></a><span class="lineno"> 196</span> </div><div class="line"><a name="l00197"></a><span class="lineno"> 197</span>  <span class="comment">// Contains ActivationWorkload</span></div><div class="line"><a name="l00198"></a><span class="lineno"> 198</span>  std::size_t found = dump.find(<span class="stringliteral">"ActivationWorkload"</span>);</div><div class="line"><a name="l00199"></a><span class="lineno"> 199</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00200"></a><span class="lineno"> 200</span> </div><div class="line"><a name="l00201"></a><span class="lineno"> 201</span>  <span class="comment">// Contains SyncMemGeneric</span></div><div class="line"><a name="l00202"></a><span class="lineno"> 202</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l00203"></a><span class="lineno"> 203</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00204"></a><span class="lineno"> 204</span> </div><div class="line"><a name="l00205"></a><span class="lineno"> 205</span>  <span class="comment">// Does not contain CopyMemGeneric</span></div><div class="line"><a name="l00206"></a><span class="lineno"> 206</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l00207"></a><span class="lineno"> 207</span>  CHECK(found == std::string::npos);</div><div class="line"><a name="l00208"></a><span class="lineno"> 208</span> </div><div class="line"><a name="l00209"></a><span class="lineno"> 209</span>  runtime->UnloadNetwork(netId);</div><div class="line"><a name="l00210"></a><span class="lineno"> 210</span> </div><div class="line"><a name="l00211"></a><span class="lineno"> 211</span>  <span class="comment">// Check output is as expected</span></div><div class="line"><a name="l00212"></a><span class="lineno"> 212</span>  <span class="comment">// Validate result by checking that the output has no negative values</span></div><div class="line"><a name="l00213"></a><span class="lineno"> 213</span>  <span class="keyword">auto</span>* outputResult = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00214"></a><span class="lineno"> 214</span>  CHECK(outputResult);</div><div class="line"><a name="l00215"></a><span class="lineno"> 215</span>  <span class="keywordflow">for</span>(<span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> i = 0; i < numElements; ++i)</div><div class="line"><a name="l00216"></a><span class="lineno"> 216</span>  {</div><div class="line"><a name="l00217"></a><span class="lineno"> 217</span>  CHECK(outputResult[i] >= 0);</div><div class="line"><a name="l00218"></a><span class="lineno"> 218</span>  }</div><div class="line"><a name="l00219"></a><span class="lineno"> 219</span> }</div><div class="line"><a name="l00220"></a><span class="lineno"> 220</span> </div><div class="line"><a name="l00221"></a><span class="lineno"> 221</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClCanBeImported"</span>)</div><div class="line"><a name="l00222"></a><span class="lineno"> 222</span> {</div><div class="line"><a name="l00223"></a><span class="lineno"> 223</span>  <a class="code" href="classarmnn_1_1_cl_import_tensor_handle_factory.xhtml">ClImportTensorHandleFactory</a> handleFactory(static_cast<MemorySourceFlags>(MemorySource::Malloc),</div><div class="line"><a name="l00224"></a><span class="lineno"> 224</span>  static_cast<MemorySourceFlags>(MemorySource::Malloc));</div><div class="line"><a name="l00225"></a><span class="lineno"> 225</span> </div><div class="line"><a name="l00226"></a><span class="lineno"> 226</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> <a class="code" href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>({ 1, 24, 16, 3 }, DataType::Float32);</div><div class="line"><a name="l00227"></a><span class="lineno"> 227</span> </div><div class="line"><a name="l00228"></a><span class="lineno"> 228</span>  <span class="comment">// create TensorHandle for memory import</span></div><div class="line"><a name="l00229"></a><span class="lineno"> 229</span>  <span class="keyword">auto</span> handle = handleFactory.CreateTensorHandle(info, DataLayout::NHWC);</div><div class="line"><a name="l00230"></a><span class="lineno"> 230</span> </div><div class="line"><a name="l00231"></a><span class="lineno"> 231</span>  <span class="comment">// Get CLtensor</span></div><div class="line"><a name="l00232"></a><span class="lineno"> 232</span>  arm_compute::CLTensor& tensor = PolymorphicDowncast<ClImportTensorHandle*>(handle.get())->GetTensor();</div><div class="line"><a name="l00233"></a><span class="lineno"> 233</span> </div><div class="line"><a name="l00234"></a><span class="lineno"> 234</span>  <span class="comment">// Allocate user memory</span></div><div class="line"><a name="l00235"></a><span class="lineno"> 235</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> totalBytes = tensor.info()->total_size();</div><div class="line"><a name="l00236"></a><span class="lineno"> 236</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00237"></a><span class="lineno"> 237</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00238"></a><span class="lineno"> 238</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l00239"></a><span class="lineno"> 239</span>  <span class="keyword">auto</span> testData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00240"></a><span class="lineno"> 240</span>  <span class="keywordtype">void</span>* alignedPtr = testData.get();</div><div class="line"><a name="l00241"></a><span class="lineno"> 241</span>  CHECK(std::align(alignment, totalBytes, alignedPtr, space));</div><div class="line"><a name="l00242"></a><span class="lineno"> 242</span> </div><div class="line"><a name="l00243"></a><span class="lineno"> 243</span>  <span class="comment">// Import memory</span></div><div class="line"><a name="l00244"></a><span class="lineno"> 244</span>  CHECK_THROWS_AS(handle->CanBeImported(alignedPtr, <a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeaec0fc0100c4fc1ce4eea230c3dc10360">armnn::MemorySource::Undefined</a>), <a class="code" href="classarmnn_1_1_memory_import_exception.xhtml">MemoryImportException</a>);</div><div class="line"><a name="l00245"></a><span class="lineno"> 245</span> </div><div class="line"><a name="l00246"></a><span class="lineno"> 246</span> }</div><div class="line"><a name="l00247"></a><span class="lineno"> 247</span> </div><div class="line"><a name="l00248"></a><span class="lineno"> 248</span> TEST_CASE(<span class="stringliteral">"ClCanBeImportedAlignedMemory"</span>)</div><div class="line"><a name="l00249"></a><span class="lineno"> 249</span> {</div><div class="line"><a name="l00250"></a><span class="lineno"> 250</span>  <a class="code" href="classarmnn_1_1_cl_import_tensor_handle_factory.xhtml">ClImportTensorHandleFactory</a> handleFactory(static_cast<MemorySourceFlags>(MemorySource::Malloc),</div><div class="line"><a name="l00251"></a><span class="lineno"> 251</span>  static_cast<MemorySourceFlags>(MemorySource::Malloc));</div><div class="line"><a name="l00252"></a><span class="lineno"> 252</span> </div><div class="line"><a name="l00253"></a><span class="lineno"> 253</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> <a class="code" href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>({ 1, 1, 1, 1 }, DataType::Float32);</div><div class="line"><a name="l00254"></a><span class="lineno"> 254</span> </div><div class="line"><a name="l00255"></a><span class="lineno"> 255</span>  <span class="comment">// create TensorHandle (Memory Managed status is irrelevant)</span></div><div class="line"><a name="l00256"></a><span class="lineno"> 256</span>  <span class="keyword">auto</span> handle = handleFactory.CreateTensorHandle(info, DataLayout::NHWC);</div><div class="line"><a name="l00257"></a><span class="lineno"> 257</span>  <span class="comment">// Get CLtensor</span></div><div class="line"><a name="l00258"></a><span class="lineno"> 258</span>  arm_compute::CLTensor& tensor = PolymorphicDowncast<ClImportTensorHandle*>(handle.get())->GetTensor();</div><div class="line"><a name="l00259"></a><span class="lineno"> 259</span> </div><div class="line"><a name="l00260"></a><span class="lineno"> 260</span>  <span class="comment">// Create an aligned buffer</span></div><div class="line"><a name="l00261"></a><span class="lineno"> 261</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> totalBytes = tensor.info()->total_size();</div><div class="line"><a name="l00262"></a><span class="lineno"> 262</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00263"></a><span class="lineno"> 263</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00264"></a><span class="lineno"> 264</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l00265"></a><span class="lineno"> 265</span>  <span class="keyword">auto</span> testData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00266"></a><span class="lineno"> 266</span>  <span class="keywordtype">void</span>* alignedPtr = testData.get();</div><div class="line"><a name="l00267"></a><span class="lineno"> 267</span>  CHECK(std::align(alignment, totalBytes, alignedPtr, space));</div><div class="line"><a name="l00268"></a><span class="lineno"> 268</span> </div><div class="line"><a name="l00269"></a><span class="lineno"> 269</span>  <span class="comment">// Check aligned buffers return true</span></div><div class="line"><a name="l00270"></a><span class="lineno"> 270</span>  CHECK(handle->CanBeImported(alignedPtr, MemorySource::Malloc) == <span class="keyword">true</span>);</div><div class="line"><a name="l00271"></a><span class="lineno"> 271</span> </div><div class="line"><a name="l00272"></a><span class="lineno"> 272</span>  <span class="comment">// Due to the nature of how GPU memory is mapped it is entirely possible for memory which is misaligned on cpu</span></div><div class="line"><a name="l00273"></a><span class="lineno"> 273</span>  <span class="comment">// to be successfully import on GPU. As such there is no way to create a misaligned pointer that will always fail.</span></div><div class="line"><a name="l00274"></a><span class="lineno"> 274</span>  <span class="comment">// Rather it will succeed on some devices and fail on others. As long as a correctly aligned buffer returns true</span></div><div class="line"><a name="l00275"></a><span class="lineno"> 275</span>  <span class="comment">// we can be confident that it will be successfully imported. All other cases will need to be handled by the user.</span></div><div class="line"><a name="l00276"></a><span class="lineno"> 276</span> }</div><div class="line"><a name="l00277"></a><span class="lineno"> 277</span> </div><div class="line"><a name="l00278"></a><span class="lineno"> 278</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClForceImportConv2dEndToEnd"</span>)</div><div class="line"><a name="l00279"></a><span class="lineno"> 279</span> {</div><div class="line"><a name="l00280"></a><span class="lineno"> 280</span>  <span class="comment">// Create runtime in which test will run</span></div><div class="line"><a name="l00281"></a><span class="lineno"> 281</span>  <a class="code" href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">IRuntime::CreationOptions</a> options;</div><div class="line"><a name="l00282"></a><span class="lineno"> 282</span>  <a class="code" href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">IRuntimePtr</a> runtime(<a class="code" href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a>(options));</div><div class="line"><a name="l00283"></a><span class="lineno"> 283</span> </div><div class="line"><a name="l00284"></a><span class="lineno"> 284</span>  <span class="comment">// build up the structure of the network</span></div><div class="line"><a name="l00285"></a><span class="lineno"> 285</span>  <a class="code" href="namespacearmnn.xhtml#ace74f6f9feb95a964a49d79458232703">INetworkPtr</a> network(INetwork::Create());</div><div class="line"><a name="l00286"></a><span class="lineno"> 286</span> </div><div class="line"><a name="l00287"></a><span class="lineno"> 287</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> inputInfo({ 1, 3, 4, 1 }, DataType::Float32);</div><div class="line"><a name="l00288"></a><span class="lineno"> 288</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> kernelInfo({ 1, 3, 3, 1 }, DataType::Float32);</div><div class="line"><a name="l00289"></a><span class="lineno"> 289</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> outputInfo({ 1, 3, 4, 1 }, DataType::Float32);</div><div class="line"><a name="l00290"></a><span class="lineno"> 290</span> </div><div class="line"><a name="l00291"></a><span class="lineno"> 291</span>  kernelInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00292"></a><span class="lineno"> 292</span> </div><div class="line"><a name="l00293"></a><span class="lineno"> 293</span>  std::vector<float> kernel =</div><div class="line"><a name="l00294"></a><span class="lineno"> 294</span>  {</div><div class="line"><a name="l00295"></a><span class="lineno"> 295</span>  4, 5, 6,</div><div class="line"><a name="l00296"></a><span class="lineno"> 296</span>  0, 0, 0,</div><div class="line"><a name="l00297"></a><span class="lineno"> 297</span>  3, 2, 1</div><div class="line"><a name="l00298"></a><span class="lineno"> 298</span>  };</div><div class="line"><a name="l00299"></a><span class="lineno"> 299</span> </div><div class="line"><a name="l00300"></a><span class="lineno"> 300</span>  <span class="keyword">const</span> std::vector<float> expectedOutput =</div><div class="line"><a name="l00301"></a><span class="lineno"> 301</span>  {</div><div class="line"><a name="l00302"></a><span class="lineno"> 302</span>  23, 41, 33, 21,</div><div class="line"><a name="l00303"></a><span class="lineno"> 303</span>  44, 65, 76, 52,</div><div class="line"><a name="l00304"></a><span class="lineno"> 304</span>  82, 85, 79, 42</div><div class="line"><a name="l00305"></a><span class="lineno"> 305</span>  };</div><div class="line"><a name="l00306"></a><span class="lineno"> 306</span> </div><div class="line"><a name="l00307"></a><span class="lineno"> 307</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = inputInfo.GetNumElements();</div><div class="line"><a name="l00308"></a><span class="lineno"> 308</span>  <span class="keywordtype">size_t</span> totalBytes = numElements * <span class="keyword">sizeof</span>(float);</div><div class="line"><a name="l00309"></a><span class="lineno"> 309</span> </div><div class="line"><a name="l00310"></a><span class="lineno"> 310</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* <span class="keyword">const</span> inputLayer = network->AddInputLayer(0, <span class="stringliteral">"input"</span>);</div><div class="line"><a name="l00311"></a><span class="lineno"> 311</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(inputLayer);</div><div class="line"><a name="l00312"></a><span class="lineno"> 312</span> </div><div class="line"><a name="l00313"></a><span class="lineno"> 313</span>  <a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a> weights(kernelInfo, kernel);</div><div class="line"><a name="l00314"></a><span class="lineno"> 314</span> </div><div class="line"><a name="l00315"></a><span class="lineno"> 315</span>  <a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml">armnn::Convolution2dDescriptor</a> convDesc2d;</div><div class="line"><a name="l00316"></a><span class="lineno"> 316</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#afe6a3377c4531315354def9023c8fdda">m_StrideX</a> = 1;</div><div class="line"><a name="l00317"></a><span class="lineno"> 317</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac1fe174bbadfb39a2b636940c2e647c8">m_StrideY</a> = 1;</div><div class="line"><a name="l00318"></a><span class="lineno"> 318</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac18546ebbebbb32fe0a03baa9bf2c600">m_PadLeft</a> = 1;</div><div class="line"><a name="l00319"></a><span class="lineno"> 319</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a002bc30e590d78cbb4f4d12171055ca7">m_PadRight</a> = 1;</div><div class="line"><a name="l00320"></a><span class="lineno"> 320</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a56b51f56cef50cdfa554258eecdab046">m_PadTop</a> = 1;</div><div class="line"><a name="l00321"></a><span class="lineno"> 321</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#aa9e49717ebdb741e8c767741647fc618">m_PadBottom</a> = 1;</div><div class="line"><a name="l00322"></a><span class="lineno"> 322</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a6089e1ca91914015777ea780a513131a">m_DataLayout</a> = DataLayout::NHWC;</div><div class="line"><a name="l00323"></a><span class="lineno"> 323</span> </div><div class="line"><a name="l00324"></a><span class="lineno"> 324</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* <span class="keyword">const</span> convLayer = network->AddConvolution2dLayer(convDesc2d, <span class="stringliteral">"conv"</span>);</div><div class="line"><a name="l00325"></a><span class="lineno"> 325</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* weightsLayer = network->AddConstantLayer(weights);</div><div class="line"><a name="l00326"></a><span class="lineno"> 326</span> </div><div class="line"><a name="l00327"></a><span class="lineno"> 327</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(convLayer);</div><div class="line"><a name="l00328"></a><span class="lineno"> 328</span> </div><div class="line"><a name="l00329"></a><span class="lineno"> 329</span>  weightsLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(weights.GetInfo());</div><div class="line"><a name="l00330"></a><span class="lineno"> 330</span>  weightsLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(1u));</div><div class="line"><a name="l00331"></a><span class="lineno"> 331</span> </div><div class="line"><a name="l00332"></a><span class="lineno"> 332</span>  inputLayer->GetOutputSlot(0).Connect(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00333"></a><span class="lineno"> 333</span>  inputLayer->GetOutputSlot(0).SetTensorInfo(inputInfo);</div><div class="line"><a name="l00334"></a><span class="lineno"> 334</span> </div><div class="line"><a name="l00335"></a><span class="lineno"> 335</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* output = network->AddOutputLayer(0, <span class="stringliteral">"output"</span>);</div><div class="line"><a name="l00336"></a><span class="lineno"> 336</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(output-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00337"></a><span class="lineno"> 337</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(outputInfo);</div><div class="line"><a name="l00338"></a><span class="lineno"> 338</span> </div><div class="line"><a name="l00339"></a><span class="lineno"> 339</span>  <span class="comment">// Optimize the network</span></div><div class="line"><a name="l00340"></a><span class="lineno"> 340</span>  <a class="code" href="structarmnn_1_1_optimizer_options.xhtml">OptimizerOptions</a> optOptions;</div><div class="line"><a name="l00341"></a><span class="lineno"> 341</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">m_ImportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00342"></a><span class="lineno"> 342</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">m_ExportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00343"></a><span class="lineno"> 343</span>  std::vector<armnn::BackendId> backends = {<a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a>};</div><div class="line"><a name="l00344"></a><span class="lineno"> 344</span>  <a class="code" href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">IOptimizedNetworkPtr</a> optNet = <a class="code" href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">Optimize</a>(*network, backends, runtime->GetDeviceSpec(), optOptions);</div><div class="line"><a name="l00345"></a><span class="lineno"> 345</span>  CHECK(optNet);</div><div class="line"><a name="l00346"></a><span class="lineno"> 346</span> </div><div class="line"><a name="l00347"></a><span class="lineno"> 347</span>  <span class="comment">// Loads it into the runtime.</span></div><div class="line"><a name="l00348"></a><span class="lineno"> 348</span>  <a class="code" href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">NetworkId</a> netId;</div><div class="line"><a name="l00349"></a><span class="lineno"> 349</span>  std::string ignoredErrorMessage;</div><div class="line"><a name="l00350"></a><span class="lineno"> 350</span>  <span class="comment">// Enable Importing</span></div><div class="line"><a name="l00351"></a><span class="lineno"> 351</span>  <a class="code" href="structarmnn_1_1_i_network_properties.xhtml">INetworkProperties</a> networkProperties(<span class="keyword">false</span>, MemorySource::Undefined, MemorySource::Undefined);</div><div class="line"><a name="l00352"></a><span class="lineno"> 352</span>  runtime->LoadNetwork(netId, std::move(optNet), ignoredErrorMessage, networkProperties);</div><div class="line"><a name="l00353"></a><span class="lineno"> 353</span> </div><div class="line"><a name="l00354"></a><span class="lineno"> 354</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l00355"></a><span class="lineno"> 355</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00356"></a><span class="lineno"> 356</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00357"></a><span class="lineno"> 357</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l00358"></a><span class="lineno"> 358</span>  <span class="keyword">auto</span> inputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00359"></a><span class="lineno"> 359</span>  <span class="keywordtype">void</span>* alignedInputPtr = inputData.get();</div><div class="line"><a name="l00360"></a><span class="lineno"> 360</span>  CHECK(std::align(alignment, totalBytes, alignedInputPtr, space));</div><div class="line"><a name="l00361"></a><span class="lineno"> 361</span> </div><div class="line"><a name="l00362"></a><span class="lineno"> 362</span>  <span class="comment">// Input with negative values</span></div><div class="line"><a name="l00363"></a><span class="lineno"> 363</span>  <span class="keyword">auto</span>* inputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedInputPtr);</div><div class="line"><a name="l00364"></a><span class="lineno"> 364</span>  inputPtr[0] = 1;</div><div class="line"><a name="l00365"></a><span class="lineno"> 365</span>  inputPtr[1] = 5;</div><div class="line"><a name="l00366"></a><span class="lineno"> 366</span>  inputPtr[2] = 2;</div><div class="line"><a name="l00367"></a><span class="lineno"> 367</span>  inputPtr[3] = 3;</div><div class="line"><a name="l00368"></a><span class="lineno"> 368</span>  inputPtr[4] = 8;</div><div class="line"><a name="l00369"></a><span class="lineno"> 369</span>  inputPtr[5] = 7;</div><div class="line"><a name="l00370"></a><span class="lineno"> 370</span>  inputPtr[6] = 3;</div><div class="line"><a name="l00371"></a><span class="lineno"> 371</span>  inputPtr[7] = 6;</div><div class="line"><a name="l00372"></a><span class="lineno"> 372</span>  inputPtr[8] = 3;</div><div class="line"><a name="l00373"></a><span class="lineno"> 373</span>  inputPtr[9] = 3;</div><div class="line"><a name="l00374"></a><span class="lineno"> 374</span>  inputPtr[10] = 9;</div><div class="line"><a name="l00375"></a><span class="lineno"> 375</span>  inputPtr[11] = 1;</div><div class="line"><a name="l00376"></a><span class="lineno"> 376</span> </div><div class="line"><a name="l00377"></a><span class="lineno"> 377</span> </div><div class="line"><a name="l00378"></a><span class="lineno"> 378</span>  <span class="keyword">auto</span> outputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00379"></a><span class="lineno"> 379</span>  <span class="keywordtype">void</span>* alignedOutputPtr = outputData.get();</div><div class="line"><a name="l00380"></a><span class="lineno"> 380</span>  CHECK(std::align(alignment, totalBytes, alignedOutputPtr, space));</div><div class="line"><a name="l00381"></a><span class="lineno"> 381</span>  <span class="keyword">auto</span>* outputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00382"></a><span class="lineno"> 382</span>  std::fill_n(outputPtr, numElements, -10.0f);</div><div class="line"><a name="l00383"></a><span class="lineno"> 383</span> </div><div class="line"><a name="l00384"></a><span class="lineno"> 384</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> inputTensorInfo = runtime->GetInputTensorInfo(netId, 0);</div><div class="line"><a name="l00385"></a><span class="lineno"> 385</span>  inputTensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00386"></a><span class="lineno"> 386</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensors</div><div class="line"><a name="l00387"></a><span class="lineno"> 387</span>  {</div><div class="line"><a name="l00388"></a><span class="lineno"> 388</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, alignedInputPtr)},</div><div class="line"><a name="l00389"></a><span class="lineno"> 389</span>  };</div><div class="line"><a name="l00390"></a><span class="lineno"> 390</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensors</div><div class="line"><a name="l00391"></a><span class="lineno"> 391</span>  {</div><div class="line"><a name="l00392"></a><span class="lineno"> 392</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), alignedOutputPtr)}</div><div class="line"><a name="l00393"></a><span class="lineno"> 393</span>  };</div><div class="line"><a name="l00394"></a><span class="lineno"> 394</span> </div><div class="line"><a name="l00395"></a><span class="lineno"> 395</span>  runtime->GetProfiler(netId)->EnableProfiling(<span class="keyword">true</span>);</div><div class="line"><a name="l00396"></a><span class="lineno"> 396</span> </div><div class="line"><a name="l00397"></a><span class="lineno"> 397</span>  INFO(<span class="stringliteral">"Run ImportInputs"</span>);</div><div class="line"><a name="l00398"></a><span class="lineno"> 398</span>  std::vector<ImportedInputId> importedInputIds =</div><div class="line"><a name="l00399"></a><span class="lineno"> 399</span>  runtime->ImportInputs(netId, inputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00400"></a><span class="lineno"> 400</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00401"></a><span class="lineno"> 401</span>  CHECK(importedInputIds.size() == 1);</div><div class="line"><a name="l00402"></a><span class="lineno"> 402</span>  std::vector<ImportedOutputId> importedOutputIds =</div><div class="line"><a name="l00403"></a><span class="lineno"> 403</span>  runtime->ImportOutputs(netId, outputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00404"></a><span class="lineno"> 404</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00405"></a><span class="lineno"> 405</span>  CHECK(importedOutputIds.size() == 1);</div><div class="line"><a name="l00406"></a><span class="lineno"> 406</span>  <span class="comment">// Do the inference</span></div><div class="line"><a name="l00407"></a><span class="lineno"> 407</span>  runtime->EnqueueWorkload(netId, <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a>(), <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a>(), importedInputIds, importedOutputIds);</div><div class="line"><a name="l00408"></a><span class="lineno"> 408</span> </div><div class="line"><a name="l00409"></a><span class="lineno"> 409</span>  <span class="comment">// Retrieve the Profiler.Print() output to get the workload execution</span></div><div class="line"><a name="l00410"></a><span class="lineno"> 410</span>  <a class="code" href="classarmnn_1_1_profiler_manager.xhtml">ProfilerManager</a>& profilerManager = <a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a>();</div><div class="line"><a name="l00411"></a><span class="lineno"> 411</span>  std::stringstream ss;</div><div class="line"><a name="l00412"></a><span class="lineno"> 412</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#a038bb767bbc6abc0ee0d9b509616b896">Print</a>(ss);;</div><div class="line"><a name="l00413"></a><span class="lineno"> 413</span>  std::string dump = ss.str();</div><div class="line"><a name="l00414"></a><span class="lineno"> 414</span> </div><div class="line"><a name="l00415"></a><span class="lineno"> 415</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l00416"></a><span class="lineno"> 416</span>  std::size_t found = dump.find(<span class="stringliteral">"Convolution2dWorkload"</span>);</div><div class="line"><a name="l00417"></a><span class="lineno"> 417</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00418"></a><span class="lineno"> 418</span> </div><div class="line"><a name="l00419"></a><span class="lineno"> 419</span>  <span class="comment">// Contains SyncMemGeneric</span></div><div class="line"><a name="l00420"></a><span class="lineno"> 420</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l00421"></a><span class="lineno"> 421</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00422"></a><span class="lineno"> 422</span> </div><div class="line"><a name="l00423"></a><span class="lineno"> 423</span>  <span class="comment">// Does not contain CopyMemGeneric</span></div><div class="line"><a name="l00424"></a><span class="lineno"> 424</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l00425"></a><span class="lineno"> 425</span>  CHECK(found == std::string::npos);</div><div class="line"><a name="l00426"></a><span class="lineno"> 426</span> </div><div class="line"><a name="l00427"></a><span class="lineno"> 427</span>  runtime->UnloadNetwork(netId);</div><div class="line"><a name="l00428"></a><span class="lineno"> 428</span> </div><div class="line"><a name="l00429"></a><span class="lineno"> 429</span>  <span class="comment">// Check output is as expected</span></div><div class="line"><a name="l00430"></a><span class="lineno"> 430</span>  <span class="comment">// Validate result by checking that the output has no negative values</span></div><div class="line"><a name="l00431"></a><span class="lineno"> 431</span>  <span class="keyword">auto</span>* outputResult = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00432"></a><span class="lineno"> 432</span>  CHECK(outputResult);</div><div class="line"><a name="l00433"></a><span class="lineno"> 433</span> </div><div class="line"><a name="l00434"></a><span class="lineno"> 434</span>  <span class="comment">// Check the output is correct</span></div><div class="line"><a name="l00435"></a><span class="lineno"> 435</span>  CHECK(std::equal(outputResult, outputResult + numElements, expectedOutput.begin(), expectedOutput.end()));</div><div class="line"><a name="l00436"></a><span class="lineno"> 436</span> }</div><div class="line"><a name="l00437"></a><span class="lineno"> 437</span> </div><div class="line"><a name="l00438"></a><span class="lineno"> 438</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClForceImportConvertFp16toFp32EndToEnd"</span>)</div><div class="line"><a name="l00439"></a><span class="lineno"> 439</span> {</div><div class="line"><a name="l00440"></a><span class="lineno"> 440</span>  <span class="keyword">using namespace </span>half_float::literal;</div><div class="line"><a name="l00441"></a><span class="lineno"> 441</span> </div><div class="line"><a name="l00442"></a><span class="lineno"> 442</span>  <span class="comment">// Create runtime in which test will run</span></div><div class="line"><a name="l00443"></a><span class="lineno"> 443</span>  <a class="code" href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">IRuntime::CreationOptions</a> options;</div><div class="line"><a name="l00444"></a><span class="lineno"> 444</span>  <a class="code" href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">IRuntimePtr</a> runtime(<a class="code" href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a>(options));</div><div class="line"><a name="l00445"></a><span class="lineno"> 445</span> </div><div class="line"><a name="l00446"></a><span class="lineno"> 446</span>  <span class="comment">// build up the structure of the network</span></div><div class="line"><a name="l00447"></a><span class="lineno"> 447</span>  <a class="code" href="classarmnn_1_1_network_impl.xhtml">NetworkImpl</a> network;</div><div class="line"><a name="l00448"></a><span class="lineno"> 448</span> </div><div class="line"><a name="l00449"></a><span class="lineno"> 449</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> inputInfo({1, 3, 2, 3}, <a class="code" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a26e6ed77470c6f2f830ecf874e6c0d55">armnn::DataType::Float16</a>);</div><div class="line"><a name="l00450"></a><span class="lineno"> 450</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> outputTensorInfo({1, 3, 2, 3}, <a class="code" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a166495adc0d0f53bee6baecc577f5204">armnn::DataType::Float32</a>);</div><div class="line"><a name="l00451"></a><span class="lineno"> 451</span> </div><div class="line"><a name="l00452"></a><span class="lineno"> 452</span>  std::vector<float> expectedOutput =</div><div class="line"><a name="l00453"></a><span class="lineno"> 453</span>  {</div><div class="line"><a name="l00454"></a><span class="lineno"> 454</span>  -37.5f, -15.2f, -8.76f, -2.0f, -1.5f, -1.3f, -0.5f, -0.4f, 0.0f,</div><div class="line"><a name="l00455"></a><span class="lineno"> 455</span>  1.0f, 0.4f, 0.5f, 1.3f, 1.5f, 2.0f, 8.76f, 15.2f, 37.5f</div><div class="line"><a name="l00456"></a><span class="lineno"> 456</span>  };</div><div class="line"><a name="l00457"></a><span class="lineno"> 457</span> </div><div class="line"><a name="l00458"></a><span class="lineno"> 458</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = inputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8846406ac37fbd2204f0be16ee05d5b7">GetNumElements</a>();</div><div class="line"><a name="l00459"></a><span class="lineno"> 459</span>  <span class="keywordtype">size_t</span> totalBytesInput = numElements * <span class="keyword">sizeof</span>(<a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>);</div><div class="line"><a name="l00460"></a><span class="lineno"> 460</span>  <span class="keywordtype">size_t</span> totalBytesOutput = numElements * <span class="keyword">sizeof</span>(float);</div><div class="line"><a name="l00461"></a><span class="lineno"> 461</span> </div><div class="line"><a name="l00462"></a><span class="lineno"> 462</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* <span class="keyword">const</span> inputLayer = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#aa6c1c42ea44777302e87ce0fad5ac510">AddInputLayer</a>(0, <span class="stringliteral">"input"</span>);</div><div class="line"><a name="l00463"></a><span class="lineno"> 463</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(inputLayer);</div><div class="line"><a name="l00464"></a><span class="lineno"> 464</span> </div><div class="line"><a name="l00465"></a><span class="lineno"> 465</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* <span class="keyword">const</span> convLayer = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#a2090bf6670b73c69309ed592068aa3af">AddConvertFp16ToFp32Layer</a>(<span class="stringliteral">"convert"</span>);</div><div class="line"><a name="l00466"></a><span class="lineno"> 466</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(convLayer);</div><div class="line"><a name="l00467"></a><span class="lineno"> 467</span> </div><div class="line"><a name="l00468"></a><span class="lineno"> 468</span>  inputLayer->GetOutputSlot(0).Connect(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00469"></a><span class="lineno"> 469</span>  inputLayer->GetOutputSlot(0).SetTensorInfo(inputInfo);</div><div class="line"><a name="l00470"></a><span class="lineno"> 470</span> </div><div class="line"><a name="l00471"></a><span class="lineno"> 471</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* output = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#af5790069aa11fd1c5bb2e17cecb06528">AddOutputLayer</a>(0, <span class="stringliteral">"output"</span>);</div><div class="line"><a name="l00472"></a><span class="lineno"> 472</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(output-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00473"></a><span class="lineno"> 473</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(outputTensorInfo);</div><div class="line"><a name="l00474"></a><span class="lineno"> 474</span> </div><div class="line"><a name="l00475"></a><span class="lineno"> 475</span>  <span class="comment">// Optimize the network</span></div><div class="line"><a name="l00476"></a><span class="lineno"> 476</span>  <a class="code" href="structarmnn_1_1_optimizer_options.xhtml">OptimizerOptions</a> optOptions;</div><div class="line"><a name="l00477"></a><span class="lineno"> 477</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">m_ImportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00478"></a><span class="lineno"> 478</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">m_ExportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00479"></a><span class="lineno"> 479</span>  std::vector<armnn::BackendId> backends = {<a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a>};</div><div class="line"><a name="l00480"></a><span class="lineno"> 480</span>  <a class="code" href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">IOptimizedNetworkPtr</a> optNet = <a class="code" href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">Optimize</a>(network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#afe0a4f719f9752a405e71878da7012ba">GetGraph</a>(), backends, runtime->GetDeviceSpec(), optOptions);</div><div class="line"><a name="l00481"></a><span class="lineno"> 481</span>  CHECK(optNet);</div><div class="line"><a name="l00482"></a><span class="lineno"> 482</span> </div><div class="line"><a name="l00483"></a><span class="lineno"> 483</span>  <span class="comment">// Loads it into the runtime.</span></div><div class="line"><a name="l00484"></a><span class="lineno"> 484</span>  <a class="code" href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">NetworkId</a> netId;</div><div class="line"><a name="l00485"></a><span class="lineno"> 485</span>  std::string ignoredErrorMessage;</div><div class="line"><a name="l00486"></a><span class="lineno"> 486</span>  <span class="comment">// Enable Importing</span></div><div class="line"><a name="l00487"></a><span class="lineno"> 487</span>  <a class="code" href="structarmnn_1_1_i_network_properties.xhtml">INetworkProperties</a> networkProperties(<span class="keyword">false</span>, MemorySource::Undefined, MemorySource::Undefined);</div><div class="line"><a name="l00488"></a><span class="lineno"> 488</span>  runtime->LoadNetwork(netId, std::move(optNet), ignoredErrorMessage, networkProperties);</div><div class="line"><a name="l00489"></a><span class="lineno"> 489</span> </div><div class="line"><a name="l00490"></a><span class="lineno"> 490</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l00491"></a><span class="lineno"> 491</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00492"></a><span class="lineno"> 492</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00493"></a><span class="lineno"> 493</span>  <span class="keywordtype">size_t</span> spaceInput = totalBytesInput + alignment + alignment;</div><div class="line"><a name="l00494"></a><span class="lineno"> 494</span>  <span class="keywordtype">size_t</span> spaceOutput = totalBytesOutput + alignment + alignment;</div><div class="line"><a name="l00495"></a><span class="lineno"> 495</span>  <span class="keyword">auto</span> inputData = std::make_unique<uint8_t[]>(spaceInput);</div><div class="line"><a name="l00496"></a><span class="lineno"> 496</span>  <span class="keywordtype">void</span>* alignedInputPtr = inputData.get();</div><div class="line"><a name="l00497"></a><span class="lineno"> 497</span>  CHECK(std::align(alignment, totalBytesInput, alignedInputPtr, spaceInput));</div><div class="line"><a name="l00498"></a><span class="lineno"> 498</span> </div><div class="line"><a name="l00499"></a><span class="lineno"> 499</span>  <span class="comment">// Input with negative values</span></div><div class="line"><a name="l00500"></a><span class="lineno"> 500</span>  <span class="keyword">auto</span>* inputPtr = <span class="keyword">reinterpret_cast<</span><a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>*<span class="keyword">></span>(alignedInputPtr);</div><div class="line"><a name="l00501"></a><span class="lineno"> 501</span>  inputPtr[0] = -37.5_h;</div><div class="line"><a name="l00502"></a><span class="lineno"> 502</span>  inputPtr[1] = -15.2_h;</div><div class="line"><a name="l00503"></a><span class="lineno"> 503</span>  inputPtr[2] = -8.76_h;</div><div class="line"><a name="l00504"></a><span class="lineno"> 504</span>  inputPtr[3] = -2.0_h;</div><div class="line"><a name="l00505"></a><span class="lineno"> 505</span>  inputPtr[4] = -1.5_h;</div><div class="line"><a name="l00506"></a><span class="lineno"> 506</span>  inputPtr[5] = -1.3_h;</div><div class="line"><a name="l00507"></a><span class="lineno"> 507</span>  inputPtr[6] = -0.5_h;</div><div class="line"><a name="l00508"></a><span class="lineno"> 508</span>  inputPtr[7] = -0.4_h;</div><div class="line"><a name="l00509"></a><span class="lineno"> 509</span>  inputPtr[8] = 0.0_h;</div><div class="line"><a name="l00510"></a><span class="lineno"> 510</span>  inputPtr[9] = 1.0_h;</div><div class="line"><a name="l00511"></a><span class="lineno"> 511</span>  inputPtr[10] = 0.4_h;</div><div class="line"><a name="l00512"></a><span class="lineno"> 512</span>  inputPtr[11] = 0.5_h;</div><div class="line"><a name="l00513"></a><span class="lineno"> 513</span>  inputPtr[12] = 1.3_h;</div><div class="line"><a name="l00514"></a><span class="lineno"> 514</span>  inputPtr[13] = 1.5_h;</div><div class="line"><a name="l00515"></a><span class="lineno"> 515</span>  inputPtr[14] = 2.0_h;</div><div class="line"><a name="l00516"></a><span class="lineno"> 516</span>  inputPtr[15] = 8.76_h;</div><div class="line"><a name="l00517"></a><span class="lineno"> 517</span>  inputPtr[16] = 15.2_h;</div><div class="line"><a name="l00518"></a><span class="lineno"> 518</span>  inputPtr[17] = 37.5_h;</div><div class="line"><a name="l00519"></a><span class="lineno"> 519</span> </div><div class="line"><a name="l00520"></a><span class="lineno"> 520</span>  <span class="keyword">auto</span> outputData = std::make_unique<uint8_t[]>(spaceOutput);</div><div class="line"><a name="l00521"></a><span class="lineno"> 521</span>  <span class="keywordtype">void</span>* alignedOutputPtr = outputData.get();</div><div class="line"><a name="l00522"></a><span class="lineno"> 522</span>  CHECK(std::align(alignment, totalBytesOutput, alignedOutputPtr, spaceOutput));</div><div class="line"><a name="l00523"></a><span class="lineno"> 523</span>  <span class="keyword">auto</span>* outputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00524"></a><span class="lineno"> 524</span>  std::fill_n(outputPtr, numElements, -10.0f);</div><div class="line"><a name="l00525"></a><span class="lineno"> 525</span> </div><div class="line"><a name="l00526"></a><span class="lineno"> 526</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> inputTensorInfo = runtime->GetInputTensorInfo(netId, 0);</div><div class="line"><a name="l00527"></a><span class="lineno"> 527</span>  inputTensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00528"></a><span class="lineno"> 528</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensors</div><div class="line"><a name="l00529"></a><span class="lineno"> 529</span>  {</div><div class="line"><a name="l00530"></a><span class="lineno"> 530</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, alignedInputPtr)},</div><div class="line"><a name="l00531"></a><span class="lineno"> 531</span>  };</div><div class="line"><a name="l00532"></a><span class="lineno"> 532</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensors</div><div class="line"><a name="l00533"></a><span class="lineno"> 533</span>  {</div><div class="line"><a name="l00534"></a><span class="lineno"> 534</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), alignedOutputPtr)}</div><div class="line"><a name="l00535"></a><span class="lineno"> 535</span>  };</div><div class="line"><a name="l00536"></a><span class="lineno"> 536</span> </div><div class="line"><a name="l00537"></a><span class="lineno"> 537</span>  runtime->GetProfiler(netId)->EnableProfiling(<span class="keyword">true</span>);</div><div class="line"><a name="l00538"></a><span class="lineno"> 538</span> </div><div class="line"><a name="l00539"></a><span class="lineno"> 539</span>  INFO(<span class="stringliteral">"Run ImportInputs"</span>);</div><div class="line"><a name="l00540"></a><span class="lineno"> 540</span>  std::vector<ImportedInputId> importedInputIds =</div><div class="line"><a name="l00541"></a><span class="lineno"> 541</span>  runtime->ImportInputs(netId, inputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00542"></a><span class="lineno"> 542</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00543"></a><span class="lineno"> 543</span>  CHECK(importedInputIds.size() == 1);</div><div class="line"><a name="l00544"></a><span class="lineno"> 544</span>  std::vector<ImportedOutputId> importedOutputIds =</div><div class="line"><a name="l00545"></a><span class="lineno"> 545</span>  runtime->ImportOutputs(netId, outputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00546"></a><span class="lineno"> 546</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00547"></a><span class="lineno"> 547</span>  CHECK(importedOutputIds.size() == 1);</div><div class="line"><a name="l00548"></a><span class="lineno"> 548</span> </div><div class="line"><a name="l00549"></a><span class="lineno"> 549</span>  <span class="comment">// Do the inference</span></div><div class="line"><a name="l00550"></a><span class="lineno"> 550</span>  runtime->EnqueueWorkload(netId, <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a>(), <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a>(), importedInputIds, importedOutputIds);</div><div class="line"><a name="l00551"></a><span class="lineno"> 551</span> </div><div class="line"><a name="l00552"></a><span class="lineno"> 552</span>  <span class="comment">// Retrieve the Profiler.Print() output to get the workload execution</span></div><div class="line"><a name="l00553"></a><span class="lineno"> 553</span>  <a class="code" href="classarmnn_1_1_profiler_manager.xhtml">ProfilerManager</a>& profilerManager = <a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a>();</div><div class="line"><a name="l00554"></a><span class="lineno"> 554</span>  std::stringstream ss;</div><div class="line"><a name="l00555"></a><span class="lineno"> 555</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#a038bb767bbc6abc0ee0d9b509616b896">Print</a>(ss);;</div><div class="line"><a name="l00556"></a><span class="lineno"> 556</span>  std::string dump = ss.str();</div><div class="line"><a name="l00557"></a><span class="lineno"> 557</span> </div><div class="line"><a name="l00558"></a><span class="lineno"> 558</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l00559"></a><span class="lineno"> 559</span>  std::size_t found = dump.find(<span class="stringliteral">"ConvertFp16ToFp32Workload"</span>);</div><div class="line"><a name="l00560"></a><span class="lineno"> 560</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00561"></a><span class="lineno"> 561</span> </div><div class="line"><a name="l00562"></a><span class="lineno"> 562</span>  <span class="comment">// Contains SyncMemGeneric</span></div><div class="line"><a name="l00563"></a><span class="lineno"> 563</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l00564"></a><span class="lineno"> 564</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00565"></a><span class="lineno"> 565</span> </div><div class="line"><a name="l00566"></a><span class="lineno"> 566</span>  <span class="comment">// Does not contain CopyMemGeneric</span></div><div class="line"><a name="l00567"></a><span class="lineno"> 567</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l00568"></a><span class="lineno"> 568</span>  CHECK(found == std::string::npos);</div><div class="line"><a name="l00569"></a><span class="lineno"> 569</span> </div><div class="line"><a name="l00570"></a><span class="lineno"> 570</span>  runtime->UnloadNetwork(netId);</div><div class="line"><a name="l00571"></a><span class="lineno"> 571</span> </div><div class="line"><a name="l00572"></a><span class="lineno"> 572</span>  <span class="comment">// Check output is as expected</span></div><div class="line"><a name="l00573"></a><span class="lineno"> 573</span>  <span class="comment">// Validate result by checking that the output has no negative values</span></div><div class="line"><a name="l00574"></a><span class="lineno"> 574</span>  <span class="keyword">auto</span>* outputResult = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00575"></a><span class="lineno"> 575</span>  CHECK(outputResult);</div><div class="line"><a name="l00576"></a><span class="lineno"> 576</span> </div><div class="line"><a name="l00577"></a><span class="lineno"> 577</span>  <span class="comment">// Check the output is correct</span></div><div class="line"><a name="l00578"></a><span class="lineno"> 578</span>  <span class="keywordflow">for</span> (<span class="keywordtype">size_t</span> i = 0; i < numElements; ++i)</div><div class="line"><a name="l00579"></a><span class="lineno"> 579</span>  {</div><div class="line"><a name="l00580"></a><span class="lineno"> 580</span>  DOCTEST_CHECK_MESSAGE(outputResult[i] == doctest::Approx(expectedOutput[i]).epsilon(0.0004),</div><div class="line"><a name="l00581"></a><span class="lineno"> 581</span>  <span class="stringliteral">"outputValue["</span> << i << <span class="stringliteral">"]: "</span> << outputResult[i] << <span class="stringliteral">" != "</span> << expectedOutput[i]);</div><div class="line"><a name="l00582"></a><span class="lineno"> 582</span>  }</div><div class="line"><a name="l00583"></a><span class="lineno"> 583</span> }</div><div class="line"><a name="l00584"></a><span class="lineno"> 584</span> </div><div class="line"><a name="l00585"></a><span class="lineno"> 585</span> </div><div class="line"><a name="l00586"></a><span class="lineno"> 586</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClForceImportConvertFp32toFp16EndToEnd"</span>)</div><div class="line"><a name="l00587"></a><span class="lineno"> 587</span> {</div><div class="line"><a name="l00588"></a><span class="lineno"> 588</span>  <span class="keyword">using namespace </span>half_float::literal;</div><div class="line"><a name="l00589"></a><span class="lineno"> 589</span> </div><div class="line"><a name="l00590"></a><span class="lineno"> 590</span>  <span class="comment">// Create runtime in which test will run</span></div><div class="line"><a name="l00591"></a><span class="lineno"> 591</span>  <a class="code" href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">IRuntime::CreationOptions</a> options;</div><div class="line"><a name="l00592"></a><span class="lineno"> 592</span>  <a class="code" href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">IRuntimePtr</a> runtime(<a class="code" href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a>(options));</div><div class="line"><a name="l00593"></a><span class="lineno"> 593</span> </div><div class="line"><a name="l00594"></a><span class="lineno"> 594</span>  <span class="comment">// build up the structure of the network</span></div><div class="line"><a name="l00595"></a><span class="lineno"> 595</span>  <a class="code" href="classarmnn_1_1_network_impl.xhtml">NetworkImpl</a> network;</div><div class="line"><a name="l00596"></a><span class="lineno"> 596</span> </div><div class="line"><a name="l00597"></a><span class="lineno"> 597</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> inputInfo({1, 3, 2, 3}, <a class="code" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a166495adc0d0f53bee6baecc577f5204">armnn::DataType::Float32</a>);</div><div class="line"><a name="l00598"></a><span class="lineno"> 598</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> outputTensorInfo({1, 3, 2, 3}, <a class="code" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a26e6ed77470c6f2f830ecf874e6c0d55">armnn::DataType::Float16</a>);</div><div class="line"><a name="l00599"></a><span class="lineno"> 599</span> </div><div class="line"><a name="l00600"></a><span class="lineno"> 600</span>  std::vector<Half> expectedOutput =</div><div class="line"><a name="l00601"></a><span class="lineno"> 601</span>  {</div><div class="line"><a name="l00602"></a><span class="lineno"> 602</span>  -37.5_h, -15.2_h, -8.76_h, -2.0_h, -1.5_h, -1.3_h, -0.5_h, -0.4_h, 0.0_h,</div><div class="line"><a name="l00603"></a><span class="lineno"> 603</span>  1.0_h, 0.4_h, 0.5_h, 1.3_h, 1.5_h, 2.0_h, 8.76_h, 15.2_h, 37.5_h</div><div class="line"><a name="l00604"></a><span class="lineno"> 604</span>  };</div><div class="line"><a name="l00605"></a><span class="lineno"> 605</span> </div><div class="line"><a name="l00606"></a><span class="lineno"> 606</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = inputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8846406ac37fbd2204f0be16ee05d5b7">GetNumElements</a>();</div><div class="line"><a name="l00607"></a><span class="lineno"> 607</span>  <span class="keywordtype">size_t</span> totalBytesInput = numElements * <span class="keyword">sizeof</span>(float);</div><div class="line"><a name="l00608"></a><span class="lineno"> 608</span>  <span class="keywordtype">size_t</span> totalBytesOutput = numElements * <span class="keyword">sizeof</span>(<a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>);</div><div class="line"><a name="l00609"></a><span class="lineno"> 609</span> </div><div class="line"><a name="l00610"></a><span class="lineno"> 610</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* <span class="keyword">const</span> inputLayer = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#aa6c1c42ea44777302e87ce0fad5ac510">AddInputLayer</a>(0, <span class="stringliteral">"input"</span>);</div><div class="line"><a name="l00611"></a><span class="lineno"> 611</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(inputLayer);</div><div class="line"><a name="l00612"></a><span class="lineno"> 612</span> </div><div class="line"><a name="l00613"></a><span class="lineno"> 613</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* <span class="keyword">const</span> convLayer = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#ab71c4df72f6587aea671acaacd6a0d9a">AddConvertFp32ToFp16Layer</a>(<span class="stringliteral">"convert"</span>);</div><div class="line"><a name="l00614"></a><span class="lineno"> 614</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(convLayer);</div><div class="line"><a name="l00615"></a><span class="lineno"> 615</span> </div><div class="line"><a name="l00616"></a><span class="lineno"> 616</span>  inputLayer->GetOutputSlot(0).Connect(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00617"></a><span class="lineno"> 617</span>  inputLayer->GetOutputSlot(0).SetTensorInfo(inputInfo);</div><div class="line"><a name="l00618"></a><span class="lineno"> 618</span> </div><div class="line"><a name="l00619"></a><span class="lineno"> 619</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* output = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#af5790069aa11fd1c5bb2e17cecb06528">AddOutputLayer</a>(0, <span class="stringliteral">"output"</span>);</div><div class="line"><a name="l00620"></a><span class="lineno"> 620</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(output-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00621"></a><span class="lineno"> 621</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(outputTensorInfo);</div><div class="line"><a name="l00622"></a><span class="lineno"> 622</span> </div><div class="line"><a name="l00623"></a><span class="lineno"> 623</span>  <span class="comment">// Optimize the network</span></div><div class="line"><a name="l00624"></a><span class="lineno"> 624</span>  <a class="code" href="structarmnn_1_1_optimizer_options.xhtml">OptimizerOptions</a> optOptions;</div><div class="line"><a name="l00625"></a><span class="lineno"> 625</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">m_ImportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00626"></a><span class="lineno"> 626</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">m_ExportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00627"></a><span class="lineno"> 627</span>  std::vector<armnn::BackendId> backends = {<a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a>};</div><div class="line"><a name="l00628"></a><span class="lineno"> 628</span>  <a class="code" href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">IOptimizedNetworkPtr</a> optNet = <a class="code" href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">Optimize</a>(network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#afe0a4f719f9752a405e71878da7012ba">GetGraph</a>(), backends, runtime->GetDeviceSpec(), optOptions);</div><div class="line"><a name="l00629"></a><span class="lineno"> 629</span>  CHECK(optNet);</div><div class="line"><a name="l00630"></a><span class="lineno"> 630</span> </div><div class="line"><a name="l00631"></a><span class="lineno"> 631</span>  <span class="comment">// Loads it into the runtime.</span></div><div class="line"><a name="l00632"></a><span class="lineno"> 632</span>  <a class="code" href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">NetworkId</a> netId;</div><div class="line"><a name="l00633"></a><span class="lineno"> 633</span>  std::string ignoredErrorMessage;</div><div class="line"><a name="l00634"></a><span class="lineno"> 634</span>  <span class="comment">// Enable Importing</span></div><div class="line"><a name="l00635"></a><span class="lineno"> 635</span>  <a class="code" href="structarmnn_1_1_i_network_properties.xhtml">INetworkProperties</a> networkProperties(<span class="keyword">false</span>, MemorySource::Undefined, MemorySource::Undefined);</div><div class="line"><a name="l00636"></a><span class="lineno"> 636</span>  runtime->LoadNetwork(netId, std::move(optNet), ignoredErrorMessage, networkProperties);</div><div class="line"><a name="l00637"></a><span class="lineno"> 637</span> </div><div class="line"><a name="l00638"></a><span class="lineno"> 638</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l00639"></a><span class="lineno"> 639</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00640"></a><span class="lineno"> 640</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00641"></a><span class="lineno"> 641</span>  <span class="keywordtype">size_t</span> spaceInput = totalBytesInput + alignment + alignment;</div><div class="line"><a name="l00642"></a><span class="lineno"> 642</span>  <span class="keywordtype">size_t</span> spaceOutput = totalBytesOutput + alignment + alignment;</div><div class="line"><a name="l00643"></a><span class="lineno"> 643</span>  <span class="keyword">auto</span> inputData = std::make_unique<uint8_t[]>(spaceInput);</div><div class="line"><a name="l00644"></a><span class="lineno"> 644</span>  <span class="keywordtype">void</span>* alignedInputPtr = inputData.get();</div><div class="line"><a name="l00645"></a><span class="lineno"> 645</span>  CHECK(std::align(alignment, totalBytesInput, alignedInputPtr, spaceInput));</div><div class="line"><a name="l00646"></a><span class="lineno"> 646</span> </div><div class="line"><a name="l00647"></a><span class="lineno"> 647</span>  <span class="comment">// Input with negative values</span></div><div class="line"><a name="l00648"></a><span class="lineno"> 648</span>  <span class="keyword">auto</span>* inputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedInputPtr);</div><div class="line"><a name="l00649"></a><span class="lineno"> 649</span>  inputPtr[0] = -37.5f;</div><div class="line"><a name="l00650"></a><span class="lineno"> 650</span>  inputPtr[1] = -15.2f;</div><div class="line"><a name="l00651"></a><span class="lineno"> 651</span>  inputPtr[2] = -8.76f;</div><div class="line"><a name="l00652"></a><span class="lineno"> 652</span>  inputPtr[3] = -2.0f;</div><div class="line"><a name="l00653"></a><span class="lineno"> 653</span>  inputPtr[4] = -1.5f;</div><div class="line"><a name="l00654"></a><span class="lineno"> 654</span>  inputPtr[5] = -1.3f;</div><div class="line"><a name="l00655"></a><span class="lineno"> 655</span>  inputPtr[6] = -0.5f;</div><div class="line"><a name="l00656"></a><span class="lineno"> 656</span>  inputPtr[7] = -0.4f;</div><div class="line"><a name="l00657"></a><span class="lineno"> 657</span>  inputPtr[8] = 0.0f;</div><div class="line"><a name="l00658"></a><span class="lineno"> 658</span>  inputPtr[9] = 1.0f;</div><div class="line"><a name="l00659"></a><span class="lineno"> 659</span>  inputPtr[10] = 0.4f;</div><div class="line"><a name="l00660"></a><span class="lineno"> 660</span>  inputPtr[11] = 0.5f;</div><div class="line"><a name="l00661"></a><span class="lineno"> 661</span>  inputPtr[12] = 1.3f;</div><div class="line"><a name="l00662"></a><span class="lineno"> 662</span>  inputPtr[13] = 1.5f;</div><div class="line"><a name="l00663"></a><span class="lineno"> 663</span>  inputPtr[14] = 2.0f;</div><div class="line"><a name="l00664"></a><span class="lineno"> 664</span>  inputPtr[15] = 8.76f;</div><div class="line"><a name="l00665"></a><span class="lineno"> 665</span>  inputPtr[16] = 15.2f;</div><div class="line"><a name="l00666"></a><span class="lineno"> 666</span>  inputPtr[17] = 37.5f;</div><div class="line"><a name="l00667"></a><span class="lineno"> 667</span> </div><div class="line"><a name="l00668"></a><span class="lineno"> 668</span>  <span class="keyword">auto</span> outputData = std::make_unique<uint8_t[]>(spaceOutput);</div><div class="line"><a name="l00669"></a><span class="lineno"> 669</span>  <span class="keywordtype">void</span>* alignedOutputPtr = outputData.get();</div><div class="line"><a name="l00670"></a><span class="lineno"> 670</span>  CHECK(std::align(alignment, totalBytesOutput, alignedOutputPtr, spaceOutput));</div><div class="line"><a name="l00671"></a><span class="lineno"> 671</span>  <span class="keyword">auto</span>* outputPtr = <span class="keyword">reinterpret_cast<</span><a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00672"></a><span class="lineno"> 672</span>  std::fill_n(outputPtr, numElements, -10.0f);</div><div class="line"><a name="l00673"></a><span class="lineno"> 673</span> </div><div class="line"><a name="l00674"></a><span class="lineno"> 674</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> inputTensorInfo = runtime->GetInputTensorInfo(netId, 0);</div><div class="line"><a name="l00675"></a><span class="lineno"> 675</span>  inputTensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00676"></a><span class="lineno"> 676</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensors</div><div class="line"><a name="l00677"></a><span class="lineno"> 677</span>  {</div><div class="line"><a name="l00678"></a><span class="lineno"> 678</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, alignedInputPtr)},</div><div class="line"><a name="l00679"></a><span class="lineno"> 679</span>  };</div><div class="line"><a name="l00680"></a><span class="lineno"> 680</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensors</div><div class="line"><a name="l00681"></a><span class="lineno"> 681</span>  {</div><div class="line"><a name="l00682"></a><span class="lineno"> 682</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), alignedOutputPtr)}</div><div class="line"><a name="l00683"></a><span class="lineno"> 683</span>  };</div><div class="line"><a name="l00684"></a><span class="lineno"> 684</span> </div><div class="line"><a name="l00685"></a><span class="lineno"> 685</span>  runtime->GetProfiler(netId)->EnableProfiling(<span class="keyword">true</span>);</div><div class="line"><a name="l00686"></a><span class="lineno"> 686</span> </div><div class="line"><a name="l00687"></a><span class="lineno"> 687</span>  INFO(<span class="stringliteral">"Run ImportInputs"</span>);</div><div class="line"><a name="l00688"></a><span class="lineno"> 688</span>  std::vector<ImportedInputId> importedInputIds =</div><div class="line"><a name="l00689"></a><span class="lineno"> 689</span>  runtime->ImportInputs(netId, inputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00690"></a><span class="lineno"> 690</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00691"></a><span class="lineno"> 691</span>  CHECK(importedInputIds.size() == 1);</div><div class="line"><a name="l00692"></a><span class="lineno"> 692</span>  std::vector<ImportedOutputId> importedOutputIds =</div><div class="line"><a name="l00693"></a><span class="lineno"> 693</span>  runtime->ImportOutputs(netId, outputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00694"></a><span class="lineno"> 694</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00695"></a><span class="lineno"> 695</span>  CHECK(importedOutputIds.size() == 1);</div><div class="line"><a name="l00696"></a><span class="lineno"> 696</span> </div><div class="line"><a name="l00697"></a><span class="lineno"> 697</span>  <span class="comment">// Do the inference</span></div><div class="line"><a name="l00698"></a><span class="lineno"> 698</span>  runtime->EnqueueWorkload(netId, <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a>(), <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a>(), importedInputIds, importedOutputIds);</div><div class="line"><a name="l00699"></a><span class="lineno"> 699</span> </div><div class="line"><a name="l00700"></a><span class="lineno"> 700</span>  <span class="comment">// Retrieve the Profiler.Print() output to get the workload execution</span></div><div class="line"><a name="l00701"></a><span class="lineno"> 701</span>  <a class="code" href="classarmnn_1_1_profiler_manager.xhtml">ProfilerManager</a>& profilerManager = <a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a>();</div><div class="line"><a name="l00702"></a><span class="lineno"> 702</span>  std::stringstream ss;</div><div class="line"><a name="l00703"></a><span class="lineno"> 703</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#a038bb767bbc6abc0ee0d9b509616b896">Print</a>(ss);;</div><div class="line"><a name="l00704"></a><span class="lineno"> 704</span>  std::string dump = ss.str();</div><div class="line"><a name="l00705"></a><span class="lineno"> 705</span> </div><div class="line"><a name="l00706"></a><span class="lineno"> 706</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l00707"></a><span class="lineno"> 707</span>  std::size_t found = dump.find(<span class="stringliteral">"ConvertFp32ToFp16Workload"</span>);</div><div class="line"><a name="l00708"></a><span class="lineno"> 708</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00709"></a><span class="lineno"> 709</span> </div><div class="line"><a name="l00710"></a><span class="lineno"> 710</span>  <span class="comment">// Contains SyncMemGeneric</span></div><div class="line"><a name="l00711"></a><span class="lineno"> 711</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l00712"></a><span class="lineno"> 712</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00713"></a><span class="lineno"> 713</span> </div><div class="line"><a name="l00714"></a><span class="lineno"> 714</span>  <span class="comment">// Does not contain CopyMemGeneric</span></div><div class="line"><a name="l00715"></a><span class="lineno"> 715</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l00716"></a><span class="lineno"> 716</span>  CHECK(found == std::string::npos);</div><div class="line"><a name="l00717"></a><span class="lineno"> 717</span> </div><div class="line"><a name="l00718"></a><span class="lineno"> 718</span>  runtime->UnloadNetwork(netId);</div><div class="line"><a name="l00719"></a><span class="lineno"> 719</span> </div><div class="line"><a name="l00720"></a><span class="lineno"> 720</span>  <span class="comment">// Check output is as expected</span></div><div class="line"><a name="l00721"></a><span class="lineno"> 721</span>  <span class="comment">// Validate result by checking that the output has no negative values</span></div><div class="line"><a name="l00722"></a><span class="lineno"> 722</span>  <span class="keyword">auto</span>* outputResult = <span class="keyword">reinterpret_cast<</span><a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00723"></a><span class="lineno"> 723</span>  CHECK(outputResult);</div><div class="line"><a name="l00724"></a><span class="lineno"> 724</span> </div><div class="line"><a name="l00725"></a><span class="lineno"> 725</span>  <span class="comment">// Check the output is correct</span></div><div class="line"><a name="l00726"></a><span class="lineno"> 726</span>  CHECK(std::equal(outputResult, outputResult + numElements, expectedOutput.begin(), expectedOutput.end()));</div><div class="line"><a name="l00727"></a><span class="lineno"> 727</span> }</div><div class="line"><a name="l00728"></a><span class="lineno"> 728</span> </div><div class="line"><a name="l00729"></a><span class="lineno"> 729</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClForceImportSimpleConvertFp32toFp16EndToEnd"</span>)</div><div class="line"><a name="l00730"></a><span class="lineno"> 730</span> {</div><div class="line"><a name="l00731"></a><span class="lineno"> 731</span>  <span class="keyword">using namespace </span>half_float::literal;</div><div class="line"><a name="l00732"></a><span class="lineno"> 732</span> </div><div class="line"><a name="l00733"></a><span class="lineno"> 733</span>  <span class="comment">// Create runtime in which test will run</span></div><div class="line"><a name="l00734"></a><span class="lineno"> 734</span>  <a class="code" href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">IRuntime::CreationOptions</a> options;</div><div class="line"><a name="l00735"></a><span class="lineno"> 735</span>  <a class="code" href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">IRuntimePtr</a> runtime(<a class="code" href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a>(options));</div><div class="line"><a name="l00736"></a><span class="lineno"> 736</span> </div><div class="line"><a name="l00737"></a><span class="lineno"> 737</span>  <span class="comment">// build up the structure of the network</span></div><div class="line"><a name="l00738"></a><span class="lineno"> 738</span>  <a class="code" href="classarmnn_1_1_network_impl.xhtml">NetworkImpl</a> network;</div><div class="line"><a name="l00739"></a><span class="lineno"> 739</span> </div><div class="line"><a name="l00740"></a><span class="lineno"> 740</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> inputInfo({1}, <a class="code" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a166495adc0d0f53bee6baecc577f5204">armnn::DataType::Float32</a>);</div><div class="line"><a name="l00741"></a><span class="lineno"> 741</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> outputTensorInfo({1}, <a class="code" href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a26e6ed77470c6f2f830ecf874e6c0d55">armnn::DataType::Float16</a>);</div><div class="line"><a name="l00742"></a><span class="lineno"> 742</span> </div><div class="line"><a name="l00743"></a><span class="lineno"> 743</span>  std::vector<Half> expectedOutput = { 1.0_h };</div><div class="line"><a name="l00744"></a><span class="lineno"> 744</span> </div><div class="line"><a name="l00745"></a><span class="lineno"> 745</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = inputInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8846406ac37fbd2204f0be16ee05d5b7">GetNumElements</a>();</div><div class="line"><a name="l00746"></a><span class="lineno"> 746</span>  <span class="keywordtype">size_t</span> totalBytesInput = numElements * <span class="keyword">sizeof</span>(float);</div><div class="line"><a name="l00747"></a><span class="lineno"> 747</span>  <span class="keywordtype">size_t</span> totalBytesOutput = numElements * <span class="keyword">sizeof</span>(<a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>);</div><div class="line"><a name="l00748"></a><span class="lineno"> 748</span> </div><div class="line"><a name="l00749"></a><span class="lineno"> 749</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* <span class="keyword">const</span> inputLayer = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#aa6c1c42ea44777302e87ce0fad5ac510">AddInputLayer</a>(0, <span class="stringliteral">"input"</span>);</div><div class="line"><a name="l00750"></a><span class="lineno"> 750</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(inputLayer);</div><div class="line"><a name="l00751"></a><span class="lineno"> 751</span> </div><div class="line"><a name="l00752"></a><span class="lineno"> 752</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* <span class="keyword">const</span> convLayer = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#ab71c4df72f6587aea671acaacd6a0d9a">AddConvertFp32ToFp16Layer</a>(<span class="stringliteral">"convert"</span>);</div><div class="line"><a name="l00753"></a><span class="lineno"> 753</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(convLayer);</div><div class="line"><a name="l00754"></a><span class="lineno"> 754</span> </div><div class="line"><a name="l00755"></a><span class="lineno"> 755</span>  inputLayer->GetOutputSlot(0).Connect(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00756"></a><span class="lineno"> 756</span>  inputLayer->GetOutputSlot(0).SetTensorInfo(inputInfo);</div><div class="line"><a name="l00757"></a><span class="lineno"> 757</span> </div><div class="line"><a name="l00758"></a><span class="lineno"> 758</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* output = network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#af5790069aa11fd1c5bb2e17cecb06528">AddOutputLayer</a>(0, <span class="stringliteral">"output"</span>);</div><div class="line"><a name="l00759"></a><span class="lineno"> 759</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(output-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00760"></a><span class="lineno"> 760</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(outputTensorInfo);</div><div class="line"><a name="l00761"></a><span class="lineno"> 761</span> </div><div class="line"><a name="l00762"></a><span class="lineno"> 762</span>  <span class="comment">// Optimize the network</span></div><div class="line"><a name="l00763"></a><span class="lineno"> 763</span>  <a class="code" href="structarmnn_1_1_optimizer_options.xhtml">OptimizerOptions</a> optOptions;</div><div class="line"><a name="l00764"></a><span class="lineno"> 764</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">m_ImportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00765"></a><span class="lineno"> 765</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">m_ExportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00766"></a><span class="lineno"> 766</span>  std::vector<armnn::BackendId> backends = {<a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a>};</div><div class="line"><a name="l00767"></a><span class="lineno"> 767</span>  <a class="code" href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">IOptimizedNetworkPtr</a> optNet = <a class="code" href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">Optimize</a>(network.<a class="code" href="classarmnn_1_1_network_impl.xhtml#afe0a4f719f9752a405e71878da7012ba">GetGraph</a>(), backends, runtime->GetDeviceSpec(), optOptions);</div><div class="line"><a name="l00768"></a><span class="lineno"> 768</span>  CHECK(optNet);</div><div class="line"><a name="l00769"></a><span class="lineno"> 769</span> </div><div class="line"><a name="l00770"></a><span class="lineno"> 770</span>  <span class="comment">// Loads it into the runtime.</span></div><div class="line"><a name="l00771"></a><span class="lineno"> 771</span>  <a class="code" href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">NetworkId</a> netId;</div><div class="line"><a name="l00772"></a><span class="lineno"> 772</span>  std::string ignoredErrorMessage;</div><div class="line"><a name="l00773"></a><span class="lineno"> 773</span>  <span class="comment">// Enable Importing</span></div><div class="line"><a name="l00774"></a><span class="lineno"> 774</span>  <a class="code" href="structarmnn_1_1_i_network_properties.xhtml">INetworkProperties</a> networkProperties(<span class="keyword">false</span>, MemorySource::Undefined, MemorySource::Undefined);</div><div class="line"><a name="l00775"></a><span class="lineno"> 775</span>  runtime->LoadNetwork(netId, std::move(optNet), ignoredErrorMessage, networkProperties);</div><div class="line"><a name="l00776"></a><span class="lineno"> 776</span> </div><div class="line"><a name="l00777"></a><span class="lineno"> 777</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l00778"></a><span class="lineno"> 778</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00779"></a><span class="lineno"> 779</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00780"></a><span class="lineno"> 780</span>  <span class="keywordtype">size_t</span> spaceInput = totalBytesInput + alignment + alignment;</div><div class="line"><a name="l00781"></a><span class="lineno"> 781</span>  <span class="keywordtype">size_t</span> spaceOutput = totalBytesOutput + alignment + alignment;</div><div class="line"><a name="l00782"></a><span class="lineno"> 782</span>  <span class="keyword">auto</span> inputData = std::make_unique<uint8_t[]>(spaceInput);</div><div class="line"><a name="l00783"></a><span class="lineno"> 783</span>  <span class="keywordtype">void</span>* alignedInputPtr = inputData.get();</div><div class="line"><a name="l00784"></a><span class="lineno"> 784</span>  CHECK(std::align(alignment, totalBytesInput, alignedInputPtr, spaceInput));</div><div class="line"><a name="l00785"></a><span class="lineno"> 785</span> </div><div class="line"><a name="l00786"></a><span class="lineno"> 786</span>  <span class="comment">// Input with negative values</span></div><div class="line"><a name="l00787"></a><span class="lineno"> 787</span>  <span class="keyword">auto</span>* inputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedInputPtr);</div><div class="line"><a name="l00788"></a><span class="lineno"> 788</span>  inputPtr[0] = 1.0f;</div><div class="line"><a name="l00789"></a><span class="lineno"> 789</span> </div><div class="line"><a name="l00790"></a><span class="lineno"> 790</span>  <span class="keyword">auto</span> outputData = std::make_unique<uint8_t[]>(spaceOutput);</div><div class="line"><a name="l00791"></a><span class="lineno"> 791</span>  <span class="keywordtype">void</span>* alignedOutputPtr = outputData.get();</div><div class="line"><a name="l00792"></a><span class="lineno"> 792</span>  CHECK(std::align(alignment, totalBytesOutput, alignedOutputPtr, spaceOutput));</div><div class="line"><a name="l00793"></a><span class="lineno"> 793</span>  <span class="keyword">auto</span>* outputPtr = <span class="keyword">reinterpret_cast<</span><a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00794"></a><span class="lineno"> 794</span>  std::fill_n(outputPtr, numElements, -10.0f);</div><div class="line"><a name="l00795"></a><span class="lineno"> 795</span> </div><div class="line"><a name="l00796"></a><span class="lineno"> 796</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> inputTensorInfo = runtime->GetInputTensorInfo(netId, 0);</div><div class="line"><a name="l00797"></a><span class="lineno"> 797</span>  inputTensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00798"></a><span class="lineno"> 798</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensors</div><div class="line"><a name="l00799"></a><span class="lineno"> 799</span>  {</div><div class="line"><a name="l00800"></a><span class="lineno"> 800</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, alignedInputPtr)},</div><div class="line"><a name="l00801"></a><span class="lineno"> 801</span>  };</div><div class="line"><a name="l00802"></a><span class="lineno"> 802</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensors</div><div class="line"><a name="l00803"></a><span class="lineno"> 803</span>  {</div><div class="line"><a name="l00804"></a><span class="lineno"> 804</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), alignedOutputPtr)}</div><div class="line"><a name="l00805"></a><span class="lineno"> 805</span>  };</div><div class="line"><a name="l00806"></a><span class="lineno"> 806</span> </div><div class="line"><a name="l00807"></a><span class="lineno"> 807</span>  runtime->GetProfiler(netId)->EnableProfiling(<span class="keyword">true</span>);</div><div class="line"><a name="l00808"></a><span class="lineno"> 808</span> </div><div class="line"><a name="l00809"></a><span class="lineno"> 809</span>  INFO(<span class="stringliteral">"Run ImportInputs"</span>);</div><div class="line"><a name="l00810"></a><span class="lineno"> 810</span>  std::vector<ImportedInputId> importedInputIds =</div><div class="line"><a name="l00811"></a><span class="lineno"> 811</span>  runtime->ImportInputs(netId, inputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00812"></a><span class="lineno"> 812</span>  CHECK(importedInputIds.size() == 1);</div><div class="line"><a name="l00813"></a><span class="lineno"> 813</span>  std::vector<ImportedOutputId> importedOutputIds =</div><div class="line"><a name="l00814"></a><span class="lineno"> 814</span>  runtime->ImportOutputs(netId, outputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00815"></a><span class="lineno"> 815</span>  CHECK(importedOutputIds.size() == 1);</div><div class="line"><a name="l00816"></a><span class="lineno"> 816</span> </div><div class="line"><a name="l00817"></a><span class="lineno"> 817</span>  <span class="comment">// Do the inference</span></div><div class="line"><a name="l00818"></a><span class="lineno"> 818</span>  runtime->EnqueueWorkload(netId, <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a>(), <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a>(), importedInputIds, importedOutputIds);</div><div class="line"><a name="l00819"></a><span class="lineno"> 819</span> </div><div class="line"><a name="l00820"></a><span class="lineno"> 820</span>  <span class="comment">// Retrieve the Profiler.Print() output to get the workload execution</span></div><div class="line"><a name="l00821"></a><span class="lineno"> 821</span>  <a class="code" href="classarmnn_1_1_profiler_manager.xhtml">ProfilerManager</a>& profilerManager = <a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a>();</div><div class="line"><a name="l00822"></a><span class="lineno"> 822</span>  std::stringstream ss;</div><div class="line"><a name="l00823"></a><span class="lineno"> 823</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#a038bb767bbc6abc0ee0d9b509616b896">Print</a>(ss);;</div><div class="line"><a name="l00824"></a><span class="lineno"> 824</span>  std::string dump = ss.str();</div><div class="line"><a name="l00825"></a><span class="lineno"> 825</span> </div><div class="line"><a name="l00826"></a><span class="lineno"> 826</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l00827"></a><span class="lineno"> 827</span>  std::size_t found = dump.find(<span class="stringliteral">"ConvertFp32ToFp16Workload"</span>);</div><div class="line"><a name="l00828"></a><span class="lineno"> 828</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00829"></a><span class="lineno"> 829</span> </div><div class="line"><a name="l00830"></a><span class="lineno"> 830</span>  <span class="comment">// Contains SyncMemGeneric</span></div><div class="line"><a name="l00831"></a><span class="lineno"> 831</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l00832"></a><span class="lineno"> 832</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00833"></a><span class="lineno"> 833</span> </div><div class="line"><a name="l00834"></a><span class="lineno"> 834</span>  <span class="comment">// Does not contain CopyMemGeneric</span></div><div class="line"><a name="l00835"></a><span class="lineno"> 835</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l00836"></a><span class="lineno"> 836</span>  CHECK(found == std::string::npos);</div><div class="line"><a name="l00837"></a><span class="lineno"> 837</span> </div><div class="line"><a name="l00838"></a><span class="lineno"> 838</span>  runtime->UnloadNetwork(netId);</div><div class="line"><a name="l00839"></a><span class="lineno"> 839</span> </div><div class="line"><a name="l00840"></a><span class="lineno"> 840</span>  <span class="comment">// Check output is as expected</span></div><div class="line"><a name="l00841"></a><span class="lineno"> 841</span>  <span class="comment">// Validate result by checking that the output has no negative values</span></div><div class="line"><a name="l00842"></a><span class="lineno"> 842</span>  <span class="keyword">auto</span>* outputResult = <span class="keyword">reinterpret_cast<</span><a class="code" href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">Half</a>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00843"></a><span class="lineno"> 843</span>  CHECK(outputResult);</div><div class="line"><a name="l00844"></a><span class="lineno"> 844</span> </div><div class="line"><a name="l00845"></a><span class="lineno"> 845</span>  <span class="comment">// Check the output is correct</span></div><div class="line"><a name="l00846"></a><span class="lineno"> 846</span>  CHECK(std::equal(outputResult, outputResult + numElements, expectedOutput.begin(), expectedOutput.end()));</div><div class="line"><a name="l00847"></a><span class="lineno"> 847</span> }</div><div class="line"><a name="l00848"></a><span class="lineno"> 848</span> </div><div class="line"><a name="l00849"></a><span class="lineno"> 849</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClForceImportRepeatedInferencesEndToEndTest"</span>)</div><div class="line"><a name="l00850"></a><span class="lineno"> 850</span> {</div><div class="line"><a name="l00851"></a><span class="lineno"> 851</span> <span class="comment">/*</span></div><div class="line"><a name="l00852"></a><span class="lineno"> 852</span> <span class="comment"> * This is a test to check the functionality of the Forced Import functionality when using repeated inferences that</span></div><div class="line"><a name="l00853"></a><span class="lineno"> 853</span> <span class="comment"> * require switching from importing to copy. For the first inference we create aligned Pointers and check they are</span></div><div class="line"><a name="l00854"></a><span class="lineno"> 854</span> <span class="comment"> * imported correctly. For the second we use similar pointers but don't use PreImporting.</span></div><div class="line"><a name="l00855"></a><span class="lineno"> 855</span> <span class="comment"> */</span></div><div class="line"><a name="l00856"></a><span class="lineno"> 856</span>  <span class="comment">// Create runtime in which test will run</span></div><div class="line"><a name="l00857"></a><span class="lineno"> 857</span>  <a class="code" href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">IRuntime::CreationOptions</a> options;</div><div class="line"><a name="l00858"></a><span class="lineno"> 858</span>  <a class="code" href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">IRuntimePtr</a> runtime(<a class="code" href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a>(options));</div><div class="line"><a name="l00859"></a><span class="lineno"> 859</span> </div><div class="line"><a name="l00860"></a><span class="lineno"> 860</span>  <span class="comment">// build up the structure of the network</span></div><div class="line"><a name="l00861"></a><span class="lineno"> 861</span>  <a class="code" href="namespacearmnn.xhtml#ace74f6f9feb95a964a49d79458232703">INetworkPtr</a> network(INetwork::Create());</div><div class="line"><a name="l00862"></a><span class="lineno"> 862</span> </div><div class="line"><a name="l00863"></a><span class="lineno"> 863</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> inputInfo({ 1, 3, 4, 1 }, DataType::Float32);</div><div class="line"><a name="l00864"></a><span class="lineno"> 864</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> kernelInfo({ 1, 3, 3, 1 }, DataType::Float32);</div><div class="line"><a name="l00865"></a><span class="lineno"> 865</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> outputInfo({ 1, 3, 4, 1 }, DataType::Float32);</div><div class="line"><a name="l00866"></a><span class="lineno"> 866</span> </div><div class="line"><a name="l00867"></a><span class="lineno"> 867</span>  kernelInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00868"></a><span class="lineno"> 868</span> </div><div class="line"><a name="l00869"></a><span class="lineno"> 869</span>  std::vector<float> kernel =</div><div class="line"><a name="l00870"></a><span class="lineno"> 870</span>  {</div><div class="line"><a name="l00871"></a><span class="lineno"> 871</span>  4, 5, 6,</div><div class="line"><a name="l00872"></a><span class="lineno"> 872</span>  0, 0, 0,</div><div class="line"><a name="l00873"></a><span class="lineno"> 873</span>  3, 2, 1</div><div class="line"><a name="l00874"></a><span class="lineno"> 874</span>  };</div><div class="line"><a name="l00875"></a><span class="lineno"> 875</span> </div><div class="line"><a name="l00876"></a><span class="lineno"> 876</span>  <span class="keyword">const</span> std::vector<float> expectedOutput =</div><div class="line"><a name="l00877"></a><span class="lineno"> 877</span>  {</div><div class="line"><a name="l00878"></a><span class="lineno"> 878</span>  23, 41, 33, 21,</div><div class="line"><a name="l00879"></a><span class="lineno"> 879</span>  44, 65, 76, 52,</div><div class="line"><a name="l00880"></a><span class="lineno"> 880</span>  82, 85, 79, 42</div><div class="line"><a name="l00881"></a><span class="lineno"> 881</span>  };</div><div class="line"><a name="l00882"></a><span class="lineno"> 882</span> </div><div class="line"><a name="l00883"></a><span class="lineno"> 883</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = inputInfo.GetNumElements();</div><div class="line"><a name="l00884"></a><span class="lineno"> 884</span>  <span class="keywordtype">size_t</span> totalBytes = numElements * <span class="keyword">sizeof</span>(float);</div><div class="line"><a name="l00885"></a><span class="lineno"> 885</span> </div><div class="line"><a name="l00886"></a><span class="lineno"> 886</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* <span class="keyword">const</span> inputLayer = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#aa6c1c42ea44777302e87ce0fad5ac510">AddInputLayer</a>(0, <span class="stringliteral">"input"</span>);</div><div class="line"><a name="l00887"></a><span class="lineno"> 887</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(inputLayer);</div><div class="line"><a name="l00888"></a><span class="lineno"> 888</span> </div><div class="line"><a name="l00889"></a><span class="lineno"> 889</span>  <a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a> weights(kernelInfo, kernel);</div><div class="line"><a name="l00890"></a><span class="lineno"> 890</span> </div><div class="line"><a name="l00891"></a><span class="lineno"> 891</span>  <a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml">armnn::Convolution2dDescriptor</a> convDesc2d;</div><div class="line"><a name="l00892"></a><span class="lineno"> 892</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#afe6a3377c4531315354def9023c8fdda">m_StrideX</a> = 1;</div><div class="line"><a name="l00893"></a><span class="lineno"> 893</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac1fe174bbadfb39a2b636940c2e647c8">m_StrideY</a> = 1;</div><div class="line"><a name="l00894"></a><span class="lineno"> 894</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac18546ebbebbb32fe0a03baa9bf2c600">m_PadLeft</a> = 1;</div><div class="line"><a name="l00895"></a><span class="lineno"> 895</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a002bc30e590d78cbb4f4d12171055ca7">m_PadRight</a> = 1;</div><div class="line"><a name="l00896"></a><span class="lineno"> 896</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a56b51f56cef50cdfa554258eecdab046">m_PadTop</a> = 1;</div><div class="line"><a name="l00897"></a><span class="lineno"> 897</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#aa9e49717ebdb741e8c767741647fc618">m_PadBottom</a> = 1;</div><div class="line"><a name="l00898"></a><span class="lineno"> 898</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a6089e1ca91914015777ea780a513131a">m_DataLayout</a> = DataLayout::NHWC;</div><div class="line"><a name="l00899"></a><span class="lineno"> 899</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* <span class="keyword">const</span> convLayer = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#a383e74ef080d4a81b8b371be4b840248">AddConvolution2dLayer</a>(convDesc2d, <span class="stringliteral">"conv"</span>);</div><div class="line"><a name="l00900"></a><span class="lineno"> 900</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(convLayer);</div><div class="line"><a name="l00901"></a><span class="lineno"> 901</span> </div><div class="line"><a name="l00902"></a><span class="lineno"> 902</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* weightsLayer = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#a1aa567f46c30960851c02847dc7b4215">AddConstantLayer</a>(weights);</div><div class="line"><a name="l00903"></a><span class="lineno"> 903</span> </div><div class="line"><a name="l00904"></a><span class="lineno"> 904</span>  weightsLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(weights.GetInfo());</div><div class="line"><a name="l00905"></a><span class="lineno"> 905</span>  weightsLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(1u));</div><div class="line"><a name="l00906"></a><span class="lineno"> 906</span> </div><div class="line"><a name="l00907"></a><span class="lineno"> 907</span>  inputLayer->GetOutputSlot(0).Connect(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00908"></a><span class="lineno"> 908</span>  inputLayer->GetOutputSlot(0).SetTensorInfo(inputInfo);</div><div class="line"><a name="l00909"></a><span class="lineno"> 909</span> </div><div class="line"><a name="l00910"></a><span class="lineno"> 910</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* output = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#af5790069aa11fd1c5bb2e17cecb06528">AddOutputLayer</a>(0, <span class="stringliteral">"output"</span>);</div><div class="line"><a name="l00911"></a><span class="lineno"> 911</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(output-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l00912"></a><span class="lineno"> 912</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(outputInfo);</div><div class="line"><a name="l00913"></a><span class="lineno"> 913</span> </div><div class="line"><a name="l00914"></a><span class="lineno"> 914</span>  <span class="comment">// Optimize the network</span></div><div class="line"><a name="l00915"></a><span class="lineno"> 915</span>  <a class="code" href="structarmnn_1_1_optimizer_options.xhtml">OptimizerOptions</a> optOptions;</div><div class="line"><a name="l00916"></a><span class="lineno"> 916</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">m_ImportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00917"></a><span class="lineno"> 917</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">m_ExportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l00918"></a><span class="lineno"> 918</span>  std::vector<armnn::BackendId> backends = {<a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a>};</div><div class="line"><a name="l00919"></a><span class="lineno"> 919</span>  <a class="code" href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">IOptimizedNetworkPtr</a> optNet = <a class="code" href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">Optimize</a>(*network, backends, runtime->GetDeviceSpec(), optOptions);</div><div class="line"><a name="l00920"></a><span class="lineno"> 920</span>  CHECK(optNet);</div><div class="line"><a name="l00921"></a><span class="lineno"> 921</span> </div><div class="line"><a name="l00922"></a><span class="lineno"> 922</span>  <span class="comment">// Loads it into the runtime.</span></div><div class="line"><a name="l00923"></a><span class="lineno"> 923</span>  <a class="code" href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">NetworkId</a> netId;</div><div class="line"><a name="l00924"></a><span class="lineno"> 924</span>  std::string ignoredErrorMessage;</div><div class="line"><a name="l00925"></a><span class="lineno"> 925</span>  <span class="comment">// Enable Importing</span></div><div class="line"><a name="l00926"></a><span class="lineno"> 926</span>  <a class="code" href="structarmnn_1_1_i_network_properties.xhtml">INetworkProperties</a> networkProperties(<span class="keyword">false</span>, MemorySource::Undefined, MemorySource::Undefined);</div><div class="line"><a name="l00927"></a><span class="lineno"> 927</span>  runtime->LoadNetwork(netId, std::move(optNet), ignoredErrorMessage, networkProperties);</div><div class="line"><a name="l00928"></a><span class="lineno"> 928</span> </div><div class="line"><a name="l00929"></a><span class="lineno"> 929</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l00930"></a><span class="lineno"> 930</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l00931"></a><span class="lineno"> 931</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l00932"></a><span class="lineno"> 932</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l00933"></a><span class="lineno"> 933</span>  <span class="keyword">auto</span> inputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00934"></a><span class="lineno"> 934</span>  <span class="keywordtype">void</span>* alignedInputPtr = inputData.get();</div><div class="line"><a name="l00935"></a><span class="lineno"> 935</span>  CHECK(std::align(alignment, totalBytes, alignedInputPtr, space));</div><div class="line"><a name="l00936"></a><span class="lineno"> 936</span> </div><div class="line"><a name="l00937"></a><span class="lineno"> 937</span>  <span class="comment">// Fill input with values</span></div><div class="line"><a name="l00938"></a><span class="lineno"> 938</span>  <span class="keyword">auto</span>* inputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedInputPtr);</div><div class="line"><a name="l00939"></a><span class="lineno"> 939</span>  inputPtr[0] = 1;</div><div class="line"><a name="l00940"></a><span class="lineno"> 940</span>  inputPtr[1] = 5;</div><div class="line"><a name="l00941"></a><span class="lineno"> 941</span>  inputPtr[2] = 2;</div><div class="line"><a name="l00942"></a><span class="lineno"> 942</span>  inputPtr[3] = 3;</div><div class="line"><a name="l00943"></a><span class="lineno"> 943</span>  inputPtr[4] = 8;</div><div class="line"><a name="l00944"></a><span class="lineno"> 944</span>  inputPtr[5] = 7;</div><div class="line"><a name="l00945"></a><span class="lineno"> 945</span>  inputPtr[6] = 3;</div><div class="line"><a name="l00946"></a><span class="lineno"> 946</span>  inputPtr[7] = 6;</div><div class="line"><a name="l00947"></a><span class="lineno"> 947</span>  inputPtr[8] = 3;</div><div class="line"><a name="l00948"></a><span class="lineno"> 948</span>  inputPtr[9] = 3;</div><div class="line"><a name="l00949"></a><span class="lineno"> 949</span>  inputPtr[10] = 9;</div><div class="line"><a name="l00950"></a><span class="lineno"> 950</span>  inputPtr[11] = 1;</div><div class="line"><a name="l00951"></a><span class="lineno"> 951</span> </div><div class="line"><a name="l00952"></a><span class="lineno"> 952</span> </div><div class="line"><a name="l00953"></a><span class="lineno"> 953</span>  <span class="keyword">auto</span> outputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l00954"></a><span class="lineno"> 954</span>  <span class="keywordtype">void</span>* alignedOutputPtr = outputData.get();</div><div class="line"><a name="l00955"></a><span class="lineno"> 955</span>  CHECK(std::align(alignment, totalBytes, alignedOutputPtr, space));</div><div class="line"><a name="l00956"></a><span class="lineno"> 956</span>  <span class="keyword">auto</span>* outputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l00957"></a><span class="lineno"> 957</span>  std::fill_n(outputPtr, numElements, -10.0f);</div><div class="line"><a name="l00958"></a><span class="lineno"> 958</span> </div><div class="line"><a name="l00959"></a><span class="lineno"> 959</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> inputTensorInfo = runtime->GetInputTensorInfo(netId, 0);</div><div class="line"><a name="l00960"></a><span class="lineno"> 960</span>  inputTensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l00961"></a><span class="lineno"> 961</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensors</div><div class="line"><a name="l00962"></a><span class="lineno"> 962</span>  {</div><div class="line"><a name="l00963"></a><span class="lineno"> 963</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, alignedInputPtr)},</div><div class="line"><a name="l00964"></a><span class="lineno"> 964</span>  };</div><div class="line"><a name="l00965"></a><span class="lineno"> 965</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensors</div><div class="line"><a name="l00966"></a><span class="lineno"> 966</span>  {</div><div class="line"><a name="l00967"></a><span class="lineno"> 967</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), alignedOutputPtr)}</div><div class="line"><a name="l00968"></a><span class="lineno"> 968</span>  };</div><div class="line"><a name="l00969"></a><span class="lineno"> 969</span> </div><div class="line"><a name="l00970"></a><span class="lineno"> 970</span>  runtime->GetProfiler(netId)->EnableProfiling(<span class="keyword">true</span>);</div><div class="line"><a name="l00971"></a><span class="lineno"> 971</span> </div><div class="line"><a name="l00972"></a><span class="lineno"> 972</span>  INFO(<span class="stringliteral">"Run ImportInputs"</span>);</div><div class="line"><a name="l00973"></a><span class="lineno"> 973</span>  std::vector<ImportedInputId> importedInputIds =</div><div class="line"><a name="l00974"></a><span class="lineno"> 974</span>  runtime->ImportInputs(netId, inputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00975"></a><span class="lineno"> 975</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00976"></a><span class="lineno"> 976</span>  CHECK(importedInputIds.size() == 1);</div><div class="line"><a name="l00977"></a><span class="lineno"> 977</span>  std::vector<ImportedOutputId> importedOutputIds =</div><div class="line"><a name="l00978"></a><span class="lineno"> 978</span>  runtime->ImportOutputs(netId, outputTensors, MemorySource::Malloc);</div><div class="line"><a name="l00979"></a><span class="lineno"> 979</span>  <span class="comment">// We expect the import to have succeeded.</span></div><div class="line"><a name="l00980"></a><span class="lineno"> 980</span>  CHECK(importedOutputIds.size() == 1);</div><div class="line"><a name="l00981"></a><span class="lineno"> 981</span> </div><div class="line"><a name="l00982"></a><span class="lineno"> 982</span>  <span class="comment">// Do the inference</span></div><div class="line"><a name="l00983"></a><span class="lineno"> 983</span>  runtime->EnqueueWorkload(netId, <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a>(), <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a>(), importedInputIds, importedOutputIds);</div><div class="line"><a name="l00984"></a><span class="lineno"> 984</span> </div><div class="line"><a name="l00985"></a><span class="lineno"> 985</span>  <span class="comment">// Retrieve the Profiler.AnalyzeEventsAndWriteResults() output to get the workload execution</span></div><div class="line"><a name="l00986"></a><span class="lineno"> 986</span>  <a class="code" href="classarmnn_1_1_profiler_manager.xhtml">ProfilerManager</a>& profilerManager = <a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a>();</div><div class="line"><a name="l00987"></a><span class="lineno"> 987</span>  std::stringstream ss;</div><div class="line"><a name="l00988"></a><span class="lineno"> 988</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#ac9f22844fb2e329ffd193f2d9a8ce336">AnalyzeEventsAndWriteResults</a>(ss);</div><div class="line"><a name="l00989"></a><span class="lineno"> 989</span>  std::string dump = ss.str();</div><div class="line"><a name="l00990"></a><span class="lineno"> 990</span> </div><div class="line"><a name="l00991"></a><span class="lineno"> 991</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l00992"></a><span class="lineno"> 992</span>  std::size_t found = dump.find(<span class="stringliteral">"Convolution2dWorkload"</span>);</div><div class="line"><a name="l00993"></a><span class="lineno"> 993</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00994"></a><span class="lineno"> 994</span> </div><div class="line"><a name="l00995"></a><span class="lineno"> 995</span>  <span class="comment">// Contains SyncMemGeneric</span></div><div class="line"><a name="l00996"></a><span class="lineno"> 996</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l00997"></a><span class="lineno"> 997</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l00998"></a><span class="lineno"> 998</span> </div><div class="line"><a name="l00999"></a><span class="lineno"> 999</span>  <span class="comment">// Does not contain CopyMemGeneric</span></div><div class="line"><a name="l01000"></a><span class="lineno"> 1000</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l01001"></a><span class="lineno"> 1001</span>  CHECK(found == std::string::npos);</div><div class="line"><a name="l01002"></a><span class="lineno"> 1002</span> </div><div class="line"><a name="l01003"></a><span class="lineno"> 1003</span>  <span class="comment">// Sync the outputs so we can read the data</span></div><div class="line"><a name="l01004"></a><span class="lineno"> 1004</span>  arm_compute::CLScheduler::get().sync();</div><div class="line"><a name="l01005"></a><span class="lineno"> 1005</span> </div><div class="line"><a name="l01006"></a><span class="lineno"> 1006</span>  <span class="comment">// Check output is as expected</span></div><div class="line"><a name="l01007"></a><span class="lineno"> 1007</span>  <span class="keyword">auto</span>* outputResult = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputPtr);</div><div class="line"><a name="l01008"></a><span class="lineno"> 1008</span>  CHECK(outputResult);</div><div class="line"><a name="l01009"></a><span class="lineno"> 1009</span>  CHECK(std::equal(outputResult, outputResult + numElements, expectedOutput.begin(), expectedOutput.end()));</div><div class="line"><a name="l01010"></a><span class="lineno"> 1010</span> </div><div class="line"><a name="l01011"></a><span class="lineno"> 1011</span>  <span class="comment">// Repeat the inference, with new tensors and without using PreImporting to force it to fall back to copying</span></div><div class="line"><a name="l01012"></a><span class="lineno"> 1012</span> </div><div class="line"><a name="l01013"></a><span class="lineno"> 1013</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l01014"></a><span class="lineno"> 1014</span>  <span class="keyword">auto</span> inputDataCopy = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l01015"></a><span class="lineno"> 1015</span>  <span class="keywordtype">void</span>* copyInputPtr = inputDataCopy.get();</div><div class="line"><a name="l01016"></a><span class="lineno"> 1016</span> </div><div class="line"><a name="l01017"></a><span class="lineno"> 1017</span>  <span class="comment">// Fill input with values</span></div><div class="line"><a name="l01018"></a><span class="lineno"> 1018</span>  <span class="keyword">auto</span>* inputCopyPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(copyInputPtr);</div><div class="line"><a name="l01019"></a><span class="lineno"> 1019</span>  inputCopyPtr[0] = 1;</div><div class="line"><a name="l01020"></a><span class="lineno"> 1020</span>  inputCopyPtr[1] = 5;</div><div class="line"><a name="l01021"></a><span class="lineno"> 1021</span>  inputCopyPtr[2] = 2;</div><div class="line"><a name="l01022"></a><span class="lineno"> 1022</span>  inputCopyPtr[3] = 3;</div><div class="line"><a name="l01023"></a><span class="lineno"> 1023</span>  inputCopyPtr[4] = 8;</div><div class="line"><a name="l01024"></a><span class="lineno"> 1024</span>  inputCopyPtr[5] = 7;</div><div class="line"><a name="l01025"></a><span class="lineno"> 1025</span>  inputCopyPtr[6] = 3;</div><div class="line"><a name="l01026"></a><span class="lineno"> 1026</span>  inputCopyPtr[7] = 6;</div><div class="line"><a name="l01027"></a><span class="lineno"> 1027</span>  inputCopyPtr[8] = 3;</div><div class="line"><a name="l01028"></a><span class="lineno"> 1028</span>  inputCopyPtr[9] = 3;</div><div class="line"><a name="l01029"></a><span class="lineno"> 1029</span>  inputCopyPtr[10] = 9;</div><div class="line"><a name="l01030"></a><span class="lineno"> 1030</span>  inputCopyPtr[11] = 1;</div><div class="line"><a name="l01031"></a><span class="lineno"> 1031</span> </div><div class="line"><a name="l01032"></a><span class="lineno"> 1032</span>  <span class="comment">// Output pre-filled with -10.0f</span></div><div class="line"><a name="l01033"></a><span class="lineno"> 1033</span>  <span class="keyword">auto</span> outputDataCopy = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l01034"></a><span class="lineno"> 1034</span>  <span class="keywordtype">void</span>* copyOutputPtr = outputDataCopy.get();</div><div class="line"><a name="l01035"></a><span class="lineno"> 1035</span>  <span class="keyword">auto</span>* outputCopyPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(copyOutputPtr);</div><div class="line"><a name="l01036"></a><span class="lineno"> 1036</span>  std::fill_n(outputCopyPtr, numElements, -10.0f);</div><div class="line"><a name="l01037"></a><span class="lineno"> 1037</span> </div><div class="line"><a name="l01038"></a><span class="lineno"> 1038</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensorsCopy</div><div class="line"><a name="l01039"></a><span class="lineno"> 1039</span>  {</div><div class="line"><a name="l01040"></a><span class="lineno"> 1040</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, copyInputPtr)},</div><div class="line"><a name="l01041"></a><span class="lineno"> 1041</span>  };</div><div class="line"><a name="l01042"></a><span class="lineno"> 1042</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensorsCopy</div><div class="line"><a name="l01043"></a><span class="lineno"> 1043</span>  {</div><div class="line"><a name="l01044"></a><span class="lineno"> 1044</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), copyOutputPtr)}</div><div class="line"><a name="l01045"></a><span class="lineno"> 1045</span>  };</div><div class="line"><a name="l01046"></a><span class="lineno"> 1046</span> </div><div class="line"><a name="l01047"></a><span class="lineno"> 1047</span>  <span class="comment">// Do the inference without any pre-imported input/output ids</span></div><div class="line"><a name="l01048"></a><span class="lineno"> 1048</span>  runtime->EnqueueWorkload(netId, inputTensorsCopy, outputTensorsCopy);</div><div class="line"><a name="l01049"></a><span class="lineno"> 1049</span>  <span class="comment">// Sync the outputs so we can read the data</span></div><div class="line"><a name="l01050"></a><span class="lineno"> 1050</span>  arm_compute::CLScheduler::get().sync();</div><div class="line"><a name="l01051"></a><span class="lineno"> 1051</span> </div><div class="line"><a name="l01052"></a><span class="lineno"> 1052</span>  <span class="comment">// Check the output is correct</span></div><div class="line"><a name="l01053"></a><span class="lineno"> 1053</span>  outputResult = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(copyOutputPtr);</div><div class="line"><a name="l01054"></a><span class="lineno"> 1054</span>  CHECK(outputResult);</div><div class="line"><a name="l01055"></a><span class="lineno"> 1055</span>  CHECK(std::equal(outputResult, outputResult + numElements, expectedOutput.begin(), expectedOutput.end()));</div><div class="line"><a name="l01056"></a><span class="lineno"> 1056</span> </div><div class="line"><a name="l01057"></a><span class="lineno"> 1057</span>  <span class="comment">// Query the profiler again, this will contain the results of both inferences</span></div><div class="line"><a name="l01058"></a><span class="lineno"> 1058</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#ac9f22844fb2e329ffd193f2d9a8ce336">AnalyzeEventsAndWriteResults</a>(ss);</div><div class="line"><a name="l01059"></a><span class="lineno"> 1059</span>  dump = ss.str();</div><div class="line"><a name="l01060"></a><span class="lineno"> 1060</span> </div><div class="line"><a name="l01061"></a><span class="lineno"> 1061</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l01062"></a><span class="lineno"> 1062</span>  found = dump.find(<span class="stringliteral">"Convolution2dWorkload"</span>);</div><div class="line"><a name="l01063"></a><span class="lineno"> 1063</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01064"></a><span class="lineno"> 1064</span> </div><div class="line"><a name="l01065"></a><span class="lineno"> 1065</span>  <span class="comment">// Should still contain the SyncMemGeneric</span></div><div class="line"><a name="l01066"></a><span class="lineno"> 1066</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l01067"></a><span class="lineno"> 1067</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01068"></a><span class="lineno"> 1068</span> </div><div class="line"><a name="l01069"></a><span class="lineno"> 1069</span>  <span class="comment">// Should now also contain a CopyMemGeneric</span></div><div class="line"><a name="l01070"></a><span class="lineno"> 1070</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l01071"></a><span class="lineno"> 1071</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01072"></a><span class="lineno"> 1072</span>  runtime->UnloadNetwork(netId);</div><div class="line"><a name="l01073"></a><span class="lineno"> 1073</span> }</div><div class="line"><a name="l01074"></a><span class="lineno"> 1074</span> </div><div class="line"><a name="l01075"></a><span class="lineno"> 1075</span> <a class="code" href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a>(<a class="code" href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixture</a>, <span class="stringliteral">"ClForceImportRepeatedInferencesInvertedEndToEndTest"</span>)</div><div class="line"><a name="l01076"></a><span class="lineno"> 1076</span> {</div><div class="line"><a name="l01077"></a><span class="lineno"> 1077</span> <span class="comment">/*</span></div><div class="line"><a name="l01078"></a><span class="lineno"> 1078</span> <span class="comment"> * This test is similar to the test above but instead of importing and then copying, we start by copying and then do</span></div><div class="line"><a name="l01079"></a><span class="lineno"> 1079</span> <span class="comment"> * the import.</span></div><div class="line"><a name="l01080"></a><span class="lineno"> 1080</span> <span class="comment"> */</span></div><div class="line"><a name="l01081"></a><span class="lineno"> 1081</span>  <span class="comment">// Create runtime in which test will run</span></div><div class="line"><a name="l01082"></a><span class="lineno"> 1082</span>  <a class="code" href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">IRuntime::CreationOptions</a> options;</div><div class="line"><a name="l01083"></a><span class="lineno"> 1083</span>  <a class="code" href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">IRuntimePtr</a> runtime(<a class="code" href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a>(options));</div><div class="line"><a name="l01084"></a><span class="lineno"> 1084</span> </div><div class="line"><a name="l01085"></a><span class="lineno"> 1085</span>  <span class="comment">// build up the structure of the network</span></div><div class="line"><a name="l01086"></a><span class="lineno"> 1086</span>  <a class="code" href="namespacearmnn.xhtml#ace74f6f9feb95a964a49d79458232703">INetworkPtr</a> network(INetwork::Create());</div><div class="line"><a name="l01087"></a><span class="lineno"> 1087</span> </div><div class="line"><a name="l01088"></a><span class="lineno"> 1088</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> inputInfo({ 1, 3, 4, 1 }, DataType::Float32);</div><div class="line"><a name="l01089"></a><span class="lineno"> 1089</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> kernelInfo({ 1, 3, 3, 1 }, DataType::Float32);</div><div class="line"><a name="l01090"></a><span class="lineno"> 1090</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a> outputInfo({ 1, 3, 4, 1 }, DataType::Float32);</div><div class="line"><a name="l01091"></a><span class="lineno"> 1091</span> </div><div class="line"><a name="l01092"></a><span class="lineno"> 1092</span>  kernelInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l01093"></a><span class="lineno"> 1093</span> </div><div class="line"><a name="l01094"></a><span class="lineno"> 1094</span>  std::vector<float> kernel =</div><div class="line"><a name="l01095"></a><span class="lineno"> 1095</span>  {</div><div class="line"><a name="l01096"></a><span class="lineno"> 1096</span>  4, 5, 6,</div><div class="line"><a name="l01097"></a><span class="lineno"> 1097</span>  0, 0, 0,</div><div class="line"><a name="l01098"></a><span class="lineno"> 1098</span>  3, 2, 1</div><div class="line"><a name="l01099"></a><span class="lineno"> 1099</span>  };</div><div class="line"><a name="l01100"></a><span class="lineno"> 1100</span> </div><div class="line"><a name="l01101"></a><span class="lineno"> 1101</span>  <span class="keyword">const</span> std::vector<float> expectedOutput =</div><div class="line"><a name="l01102"></a><span class="lineno"> 1102</span>  {</div><div class="line"><a name="l01103"></a><span class="lineno"> 1103</span>  23, 41, 33, 21,</div><div class="line"><a name="l01104"></a><span class="lineno"> 1104</span>  44, 65, 76, 52,</div><div class="line"><a name="l01105"></a><span class="lineno"> 1105</span>  82, 85, 79, 42</div><div class="line"><a name="l01106"></a><span class="lineno"> 1106</span>  };</div><div class="line"><a name="l01107"></a><span class="lineno"> 1107</span> </div><div class="line"><a name="l01108"></a><span class="lineno"> 1108</span>  <span class="keywordtype">unsigned</span> <span class="keywordtype">int</span> numElements = inputInfo.GetNumElements();</div><div class="line"><a name="l01109"></a><span class="lineno"> 1109</span>  <span class="keywordtype">size_t</span> totalBytes = numElements * <span class="keyword">sizeof</span>(float);</div><div class="line"><a name="l01110"></a><span class="lineno"> 1110</span> </div><div class="line"><a name="l01111"></a><span class="lineno"> 1111</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* <span class="keyword">const</span> inputLayer = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#aa6c1c42ea44777302e87ce0fad5ac510">AddInputLayer</a>(0, <span class="stringliteral">"input"</span>);</div><div class="line"><a name="l01112"></a><span class="lineno"> 1112</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(inputLayer);</div><div class="line"><a name="l01113"></a><span class="lineno"> 1113</span> </div><div class="line"><a name="l01114"></a><span class="lineno"> 1114</span>  <a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a> weights(kernelInfo, kernel);</div><div class="line"><a name="l01115"></a><span class="lineno"> 1115</span> </div><div class="line"><a name="l01116"></a><span class="lineno"> 1116</span>  <a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml">armnn::Convolution2dDescriptor</a> convDesc2d;</div><div class="line"><a name="l01117"></a><span class="lineno"> 1117</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#afe6a3377c4531315354def9023c8fdda">m_StrideX</a> = 1;</div><div class="line"><a name="l01118"></a><span class="lineno"> 1118</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac1fe174bbadfb39a2b636940c2e647c8">m_StrideY</a> = 1;</div><div class="line"><a name="l01119"></a><span class="lineno"> 1119</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac18546ebbebbb32fe0a03baa9bf2c600">m_PadLeft</a> = 1;</div><div class="line"><a name="l01120"></a><span class="lineno"> 1120</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a002bc30e590d78cbb4f4d12171055ca7">m_PadRight</a> = 1;</div><div class="line"><a name="l01121"></a><span class="lineno"> 1121</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a56b51f56cef50cdfa554258eecdab046">m_PadTop</a> = 1;</div><div class="line"><a name="l01122"></a><span class="lineno"> 1122</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#aa9e49717ebdb741e8c767741647fc618">m_PadBottom</a> = 1;</div><div class="line"><a name="l01123"></a><span class="lineno"> 1123</span>  convDesc2d.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.xhtml#a6089e1ca91914015777ea780a513131a">m_DataLayout</a> = DataLayout::NHWC;</div><div class="line"><a name="l01124"></a><span class="lineno"> 1124</span> </div><div class="line"><a name="l01125"></a><span class="lineno"> 1125</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* <span class="keyword">const</span> convLayer = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#a383e74ef080d4a81b8b371be4b840248">AddConvolution2dLayer</a>(convDesc2d, <span class="stringliteral">"conv"</span>);</div><div class="line"><a name="l01126"></a><span class="lineno"> 1126</span>  <a class="code" href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a>(convLayer);</div><div class="line"><a name="l01127"></a><span class="lineno"> 1127</span> </div><div class="line"><a name="l01128"></a><span class="lineno"> 1128</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a>* weightsLayer = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#a1aa567f46c30960851c02847dc7b4215">AddConstantLayer</a>(weights);</div><div class="line"><a name="l01129"></a><span class="lineno"> 1129</span> </div><div class="line"><a name="l01130"></a><span class="lineno"> 1130</span>  weightsLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(weights.GetInfo());</div><div class="line"><a name="l01131"></a><span class="lineno"> 1131</span>  weightsLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(1u));</div><div class="line"><a name="l01132"></a><span class="lineno"> 1132</span> </div><div class="line"><a name="l01133"></a><span class="lineno"> 1133</span>  inputLayer->GetOutputSlot(0).Connect(convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l01134"></a><span class="lineno"> 1134</span>  inputLayer->GetOutputSlot(0).SetTensorInfo(inputInfo);</div><div class="line"><a name="l01135"></a><span class="lineno"> 1135</span> </div><div class="line"><a name="l01136"></a><span class="lineno"> 1136</span>  <a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml">IConnectableLayer</a>* output = network-><a class="code" href="classarmnn_1_1_network_impl.xhtml#af5790069aa11fd1c5bb2e17cecb06528">AddOutputLayer</a>(0, <span class="stringliteral">"output"</span>);</div><div class="line"><a name="l01137"></a><span class="lineno"> 1137</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">Connect</a>(output-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">GetInputSlot</a>(0));</div><div class="line"><a name="l01138"></a><span class="lineno"> 1138</span>  convLayer-><a class="code" href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">GetOutputSlot</a>(0).<a class="code" href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">SetTensorInfo</a>(outputInfo);</div><div class="line"><a name="l01139"></a><span class="lineno"> 1139</span> </div><div class="line"><a name="l01140"></a><span class="lineno"> 1140</span>  <span class="comment">// Optimize the network</span></div><div class="line"><a name="l01141"></a><span class="lineno"> 1141</span>  <a class="code" href="structarmnn_1_1_optimizer_options.xhtml">OptimizerOptions</a> optOptions;</div><div class="line"><a name="l01142"></a><span class="lineno"> 1142</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">m_ImportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l01143"></a><span class="lineno"> 1143</span>  optOptions.<a class="code" href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">m_ExportEnabled</a> = <span class="keyword">false</span>;</div><div class="line"><a name="l01144"></a><span class="lineno"> 1144</span>  std::vector<armnn::BackendId> backends = {<a class="code" href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a>};</div><div class="line"><a name="l01145"></a><span class="lineno"> 1145</span>  <a class="code" href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">IOptimizedNetworkPtr</a> optNet = <a class="code" href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">Optimize</a>(*network, backends, runtime->GetDeviceSpec(), optOptions);</div><div class="line"><a name="l01146"></a><span class="lineno"> 1146</span>  CHECK(optNet);</div><div class="line"><a name="l01147"></a><span class="lineno"> 1147</span> </div><div class="line"><a name="l01148"></a><span class="lineno"> 1148</span>  <span class="comment">// Loads it into the runtime.</span></div><div class="line"><a name="l01149"></a><span class="lineno"> 1149</span>  <a class="code" href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">NetworkId</a> netId;</div><div class="line"><a name="l01150"></a><span class="lineno"> 1150</span>  std::string ignoredErrorMessage;</div><div class="line"><a name="l01151"></a><span class="lineno"> 1151</span>  <span class="comment">// Enable Importing</span></div><div class="line"><a name="l01152"></a><span class="lineno"> 1152</span>  <a class="code" href="structarmnn_1_1_i_network_properties.xhtml">INetworkProperties</a> networkProperties(<span class="keyword">false</span>, MemorySource::Undefined, MemorySource::Undefined);</div><div class="line"><a name="l01153"></a><span class="lineno"> 1153</span>  runtime->LoadNetwork(netId, std::move(optNet), ignoredErrorMessage, networkProperties);</div><div class="line"><a name="l01154"></a><span class="lineno"> 1154</span> </div><div class="line"><a name="l01155"></a><span class="lineno"> 1155</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l01156"></a><span class="lineno"> 1156</span>  <span class="keyword">const</span> <span class="keywordtype">size_t</span> alignment =</div><div class="line"><a name="l01157"></a><span class="lineno"> 1157</span>  arm_compute::CLKernelLibrary::get().get_device().getInfo<CL_DEVICE_GLOBAL_MEM_CACHELINE_SIZE>();</div><div class="line"><a name="l01158"></a><span class="lineno"> 1158</span>  <span class="keywordtype">size_t</span> space = totalBytes + alignment + alignment;</div><div class="line"><a name="l01159"></a><span class="lineno"> 1159</span>  <span class="keyword">auto</span> inputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l01160"></a><span class="lineno"> 1160</span>  <span class="keywordtype">void</span>* copyInputPtr = inputData.get();</div><div class="line"><a name="l01161"></a><span class="lineno"> 1161</span> </div><div class="line"><a name="l01162"></a><span class="lineno"> 1162</span>  <span class="comment">// Fill input with values</span></div><div class="line"><a name="l01163"></a><span class="lineno"> 1163</span>  <span class="keyword">auto</span>* inputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(copyInputPtr);</div><div class="line"><a name="l01164"></a><span class="lineno"> 1164</span>  inputPtr[0] = 1;</div><div class="line"><a name="l01165"></a><span class="lineno"> 1165</span>  inputPtr[1] = 5;</div><div class="line"><a name="l01166"></a><span class="lineno"> 1166</span>  inputPtr[2] = 2;</div><div class="line"><a name="l01167"></a><span class="lineno"> 1167</span>  inputPtr[3] = 3;</div><div class="line"><a name="l01168"></a><span class="lineno"> 1168</span>  inputPtr[4] = 8;</div><div class="line"><a name="l01169"></a><span class="lineno"> 1169</span>  inputPtr[5] = 7;</div><div class="line"><a name="l01170"></a><span class="lineno"> 1170</span>  inputPtr[6] = 3;</div><div class="line"><a name="l01171"></a><span class="lineno"> 1171</span>  inputPtr[7] = 6;</div><div class="line"><a name="l01172"></a><span class="lineno"> 1172</span>  inputPtr[8] = 3;</div><div class="line"><a name="l01173"></a><span class="lineno"> 1173</span>  inputPtr[9] = 3;</div><div class="line"><a name="l01174"></a><span class="lineno"> 1174</span>  inputPtr[10] = 9;</div><div class="line"><a name="l01175"></a><span class="lineno"> 1175</span>  inputPtr[11] = 1;</div><div class="line"><a name="l01176"></a><span class="lineno"> 1176</span> </div><div class="line"><a name="l01177"></a><span class="lineno"> 1177</span>  <span class="comment">// Create output buffer and fill it with -10.0f</span></div><div class="line"><a name="l01178"></a><span class="lineno"> 1178</span>  <span class="keyword">auto</span> outputData = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l01179"></a><span class="lineno"> 1179</span>  <span class="keywordtype">void</span>* copyOutputPtr = outputData.get();</div><div class="line"><a name="l01180"></a><span class="lineno"> 1180</span>  <span class="keyword">auto</span>* outputPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(copyOutputPtr);</div><div class="line"><a name="l01181"></a><span class="lineno"> 1181</span>  std::fill_n(outputPtr, numElements, -10.0f);</div><div class="line"><a name="l01182"></a><span class="lineno"> 1182</span> </div><div class="line"><a name="l01183"></a><span class="lineno"> 1183</span>  <a class="code" href="classarmnn_1_1_tensor_info.xhtml">TensorInfo</a> inputTensorInfo = runtime->GetInputTensorInfo(netId, 0);</div><div class="line"><a name="l01184"></a><span class="lineno"> 1184</span>  inputTensorInfo.<a class="code" href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">SetConstant</a>(<span class="keyword">true</span>);</div><div class="line"><a name="l01185"></a><span class="lineno"> 1185</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensors</div><div class="line"><a name="l01186"></a><span class="lineno"> 1186</span>  {</div><div class="line"><a name="l01187"></a><span class="lineno"> 1187</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, copyInputPtr)},</div><div class="line"><a name="l01188"></a><span class="lineno"> 1188</span>  };</div><div class="line"><a name="l01189"></a><span class="lineno"> 1189</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensors</div><div class="line"><a name="l01190"></a><span class="lineno"> 1190</span>  {</div><div class="line"><a name="l01191"></a><span class="lineno"> 1191</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), copyOutputPtr)}</div><div class="line"><a name="l01192"></a><span class="lineno"> 1192</span>  };</div><div class="line"><a name="l01193"></a><span class="lineno"> 1193</span> </div><div class="line"><a name="l01194"></a><span class="lineno"> 1194</span>  runtime->GetProfiler(netId)->EnableProfiling(<span class="keyword">true</span>);</div><div class="line"><a name="l01195"></a><span class="lineno"> 1195</span> </div><div class="line"><a name="l01196"></a><span class="lineno"> 1196</span>  <span class="comment">// Do the inference without any pre-imported inputs/outputs</span></div><div class="line"><a name="l01197"></a><span class="lineno"> 1197</span>  runtime->EnqueueWorkload(netId, inputTensors, outputTensors);</div><div class="line"><a name="l01198"></a><span class="lineno"> 1198</span> </div><div class="line"><a name="l01199"></a><span class="lineno"> 1199</span>  <span class="comment">// Retrieve the Profiler.AnalyzeEventsAndWriteResults() output to get the workload execution</span></div><div class="line"><a name="l01200"></a><span class="lineno"> 1200</span>  <a class="code" href="classarmnn_1_1_profiler_manager.xhtml">ProfilerManager</a>& profilerManager = <a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a>();</div><div class="line"><a name="l01201"></a><span class="lineno"> 1201</span>  std::stringstream ss;</div><div class="line"><a name="l01202"></a><span class="lineno"> 1202</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#ac9f22844fb2e329ffd193f2d9a8ce336">AnalyzeEventsAndWriteResults</a>(ss);</div><div class="line"><a name="l01203"></a><span class="lineno"> 1203</span>  std::string dump = ss.str();</div><div class="line"><a name="l01204"></a><span class="lineno"> 1204</span> </div><div class="line"><a name="l01205"></a><span class="lineno"> 1205</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l01206"></a><span class="lineno"> 1206</span>  std::size_t found = dump.find(<span class="stringliteral">"Convolution2dWorkload"</span>);</div><div class="line"><a name="l01207"></a><span class="lineno"> 1207</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01208"></a><span class="lineno"> 1208</span> </div><div class="line"><a name="l01209"></a><span class="lineno"> 1209</span>  <span class="comment">// Does not contain SyncMemGeneric</span></div><div class="line"><a name="l01210"></a><span class="lineno"> 1210</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l01211"></a><span class="lineno"> 1211</span>  CHECK(found == std::string::npos);</div><div class="line"><a name="l01212"></a><span class="lineno"> 1212</span> </div><div class="line"><a name="l01213"></a><span class="lineno"> 1213</span>  <span class="comment">// Does contain CopyMemGeneric</span></div><div class="line"><a name="l01214"></a><span class="lineno"> 1214</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l01215"></a><span class="lineno"> 1215</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01216"></a><span class="lineno"> 1216</span> </div><div class="line"><a name="l01217"></a><span class="lineno"> 1217</span>  <span class="comment">// Sync the outputs so we can read the data</span></div><div class="line"><a name="l01218"></a><span class="lineno"> 1218</span>  arm_compute::CLScheduler::get().sync();</div><div class="line"><a name="l01219"></a><span class="lineno"> 1219</span> </div><div class="line"><a name="l01220"></a><span class="lineno"> 1220</span>  <span class="comment">// Check output is as expected</span></div><div class="line"><a name="l01221"></a><span class="lineno"> 1221</span>  <span class="keyword">auto</span>* outputResult = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(copyOutputPtr);</div><div class="line"><a name="l01222"></a><span class="lineno"> 1222</span>  CHECK(outputResult);</div><div class="line"><a name="l01223"></a><span class="lineno"> 1223</span>  CHECK(std::equal(outputResult, outputResult + numElements, expectedOutput.begin(), expectedOutput.end()));</div><div class="line"><a name="l01224"></a><span class="lineno"> 1224</span> </div><div class="line"><a name="l01225"></a><span class="lineno"> 1225</span>  <span class="comment">// Repeat the inference, with new tensors and while using pre-importing to force it to import</span></div><div class="line"><a name="l01226"></a><span class="lineno"> 1226</span> </div><div class="line"><a name="l01227"></a><span class="lineno"> 1227</span>  <span class="comment">// Creates structures for input & output</span></div><div class="line"><a name="l01228"></a><span class="lineno"> 1228</span>  <span class="keyword">auto</span> inputDataImport = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l01229"></a><span class="lineno"> 1229</span>  <span class="keywordtype">void</span>* alignedInputImportPtr = inputDataImport.get();</div><div class="line"><a name="l01230"></a><span class="lineno"> 1230</span>  CHECK(std::align(alignment, totalBytes, alignedInputImportPtr, space));</div><div class="line"><a name="l01231"></a><span class="lineno"> 1231</span> </div><div class="line"><a name="l01232"></a><span class="lineno"> 1232</span>  <span class="comment">// Fill input with values</span></div><div class="line"><a name="l01233"></a><span class="lineno"> 1233</span>  <span class="keyword">auto</span>* inputImportPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedInputImportPtr);</div><div class="line"><a name="l01234"></a><span class="lineno"> 1234</span>  inputImportPtr[0] = 1;</div><div class="line"><a name="l01235"></a><span class="lineno"> 1235</span>  inputImportPtr[1] = 5;</div><div class="line"><a name="l01236"></a><span class="lineno"> 1236</span>  inputImportPtr[2] = 2;</div><div class="line"><a name="l01237"></a><span class="lineno"> 1237</span>  inputImportPtr[3] = 3;</div><div class="line"><a name="l01238"></a><span class="lineno"> 1238</span>  inputImportPtr[4] = 8;</div><div class="line"><a name="l01239"></a><span class="lineno"> 1239</span>  inputImportPtr[5] = 7;</div><div class="line"><a name="l01240"></a><span class="lineno"> 1240</span>  inputImportPtr[6] = 3;</div><div class="line"><a name="l01241"></a><span class="lineno"> 1241</span>  inputImportPtr[7] = 6;</div><div class="line"><a name="l01242"></a><span class="lineno"> 1242</span>  inputImportPtr[8] = 3;</div><div class="line"><a name="l01243"></a><span class="lineno"> 1243</span>  inputImportPtr[9] = 3;</div><div class="line"><a name="l01244"></a><span class="lineno"> 1244</span>  inputImportPtr[10] = 9;</div><div class="line"><a name="l01245"></a><span class="lineno"> 1245</span>  inputImportPtr[11] = 1;</div><div class="line"><a name="l01246"></a><span class="lineno"> 1246</span> </div><div class="line"><a name="l01247"></a><span class="lineno"> 1247</span>  <span class="comment">// Output pre-filled with -10.0f</span></div><div class="line"><a name="l01248"></a><span class="lineno"> 1248</span>  <span class="keyword">auto</span> outputDataImport = std::make_unique<uint8_t[]>(space);</div><div class="line"><a name="l01249"></a><span class="lineno"> 1249</span>  <span class="keywordtype">void</span>* alignedOutputImportPtr = outputDataImport.get();</div><div class="line"><a name="l01250"></a><span class="lineno"> 1250</span>  CHECK(std::align(alignment, totalBytes, alignedOutputImportPtr, space));</div><div class="line"><a name="l01251"></a><span class="lineno"> 1251</span>  <span class="keyword">auto</span>* outputImportPtr = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputImportPtr);</div><div class="line"><a name="l01252"></a><span class="lineno"> 1252</span>  std::fill_n(outputImportPtr, numElements, -10.0f);</div><div class="line"><a name="l01253"></a><span class="lineno"> 1253</span> </div><div class="line"><a name="l01254"></a><span class="lineno"> 1254</span>  <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a> inputTensorsImport</div><div class="line"><a name="l01255"></a><span class="lineno"> 1255</span>  {</div><div class="line"><a name="l01256"></a><span class="lineno"> 1256</span>  {0,<a class="code" href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a>(inputTensorInfo, alignedInputImportPtr)},</div><div class="line"><a name="l01257"></a><span class="lineno"> 1257</span>  };</div><div class="line"><a name="l01258"></a><span class="lineno"> 1258</span>  <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a> outputTensorsImport</div><div class="line"><a name="l01259"></a><span class="lineno"> 1259</span>  {</div><div class="line"><a name="l01260"></a><span class="lineno"> 1260</span>  {0,<a class="code" href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a>(runtime->GetOutputTensorInfo(netId, 0), alignedOutputImportPtr)}</div><div class="line"><a name="l01261"></a><span class="lineno"> 1261</span>  };</div><div class="line"><a name="l01262"></a><span class="lineno"> 1262</span> </div><div class="line"><a name="l01263"></a><span class="lineno"> 1263</span>  INFO(<span class="stringliteral">"Run ImportInputs"</span>);</div><div class="line"><a name="l01264"></a><span class="lineno"> 1264</span>  std::vector<ImportedInputId> importedInputIds =</div><div class="line"><a name="l01265"></a><span class="lineno"> 1265</span>  runtime->ImportInputs(netId, inputTensorsImport, MemorySource::Malloc);</div><div class="line"><a name="l01266"></a><span class="lineno"> 1266</span>  CHECK(importedInputIds.size() == 1);</div><div class="line"><a name="l01267"></a><span class="lineno"> 1267</span>  std::vector<ImportedOutputId> importedOutputIds =</div><div class="line"><a name="l01268"></a><span class="lineno"> 1268</span>  runtime->ImportOutputs(netId, outputTensorsImport, MemorySource::Malloc);</div><div class="line"><a name="l01269"></a><span class="lineno"> 1269</span>  CHECK(importedOutputIds.size() == 1);</div><div class="line"><a name="l01270"></a><span class="lineno"> 1270</span> </div><div class="line"><a name="l01271"></a><span class="lineno"> 1271</span>  <span class="comment">// Do the inference with pre-imported inputs/outputs</span></div><div class="line"><a name="l01272"></a><span class="lineno"> 1272</span>  runtime->EnqueueWorkload(netId, <a class="code" href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">InputTensors</a>(), <a class="code" href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">OutputTensors</a>(), importedInputIds, importedOutputIds);</div><div class="line"><a name="l01273"></a><span class="lineno"> 1273</span>  <span class="comment">// Sync the outputs so we can read the data</span></div><div class="line"><a name="l01274"></a><span class="lineno"> 1274</span>  arm_compute::CLScheduler::get().sync();</div><div class="line"><a name="l01275"></a><span class="lineno"> 1275</span> </div><div class="line"><a name="l01276"></a><span class="lineno"> 1276</span>  <span class="comment">// Check the output is correct</span></div><div class="line"><a name="l01277"></a><span class="lineno"> 1277</span>  outputResult = <span class="keyword">reinterpret_cast<</span><span class="keywordtype">float</span>*<span class="keyword">></span>(alignedOutputImportPtr);</div><div class="line"><a name="l01278"></a><span class="lineno"> 1278</span>  CHECK(outputResult);</div><div class="line"><a name="l01279"></a><span class="lineno"> 1279</span>  CHECK(std::equal(outputResult, outputResult + numElements, expectedOutput.begin(), expectedOutput.end()));</div><div class="line"><a name="l01280"></a><span class="lineno"> 1280</span> </div><div class="line"><a name="l01281"></a><span class="lineno"> 1281</span> </div><div class="line"><a name="l01282"></a><span class="lineno"> 1282</span>  <span class="comment">// Query the profiler again, this will contain the results of both inferences</span></div><div class="line"><a name="l01283"></a><span class="lineno"> 1283</span>  profilerManager.<a class="code" href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">GetProfiler</a>()-><a class="code" href="classarmnn_1_1_i_profiler.xhtml#ac9f22844fb2e329ffd193f2d9a8ce336">AnalyzeEventsAndWriteResults</a>(ss);</div><div class="line"><a name="l01284"></a><span class="lineno"> 1284</span>  dump = ss.str();</div><div class="line"><a name="l01285"></a><span class="lineno"> 1285</span> </div><div class="line"><a name="l01286"></a><span class="lineno"> 1286</span>  <span class="comment">// Contains Convolution2dWorkload</span></div><div class="line"><a name="l01287"></a><span class="lineno"> 1287</span>  found = dump.find(<span class="stringliteral">"Convolution2dWorkload"</span>);</div><div class="line"><a name="l01288"></a><span class="lineno"> 1288</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01289"></a><span class="lineno"> 1289</span> </div><div class="line"><a name="l01290"></a><span class="lineno"> 1290</span>  <span class="comment">// Should now contain the SyncMemGeneric</span></div><div class="line"><a name="l01291"></a><span class="lineno"> 1291</span>  found = dump.find(<span class="stringliteral">"SyncMemGeneric"</span>);</div><div class="line"><a name="l01292"></a><span class="lineno"> 1292</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01293"></a><span class="lineno"> 1293</span> </div><div class="line"><a name="l01294"></a><span class="lineno"> 1294</span>  <span class="comment">// Should still contain a CopyMemGeneric from the first inference</span></div><div class="line"><a name="l01295"></a><span class="lineno"> 1295</span>  found = dump.find(<span class="stringliteral">"CopyMemGeneric"</span>);</div><div class="line"><a name="l01296"></a><span class="lineno"> 1296</span>  CHECK(found != std::string::npos);</div><div class="line"><a name="l01297"></a><span class="lineno"> 1297</span>  runtime->UnloadNetwork(netId);</div><div class="line"><a name="l01298"></a><span class="lineno"> 1298</span> }</div><div class="line"><a name="l01299"></a><span class="lineno"> 1299</span> </div><div class="line"><a name="l01300"></a><span class="lineno"> 1300</span> }</div><div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml_aa9e49717ebdb741e8c767741647fc618"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml#aa9e49717ebdb741e8c767741647fc618">armnn::Convolution2dDescriptor::m_PadBottom</a></div><div class="ttdeci">uint32_t m_PadBottom</div><div class="ttdoc">Padding bottom value in the height dimension. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00533">Descriptors.hpp:533</a></div></div> +<div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml_a6089e1ca91914015777ea780a513131a"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml#a6089e1ca91914015777ea780a513131a">armnn::Convolution2dDescriptor::m_DataLayout</a></div><div class="ttdeci">DataLayout m_DataLayout</div><div class="ttdoc">The data layout to be used (NCHW, NHWC). </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00545">Descriptors.hpp:545</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_runtime_xhtml_ad44ecd3700748dc30dc4bbe34ba5bde7"><div class="ttname"><a href="classarmnn_1_1_i_runtime.xhtml#ad44ecd3700748dc30dc4bbe34ba5bde7">armnn::IRuntime::Create</a></div><div class="ttdeci">static IRuntimePtr Create(const CreationOptions &options)</div><div class="ttdef"><b>Definition:</b> <a href="_runtime_8cpp_source.xhtml#l00049">Runtime.cpp:49</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_connectable_layer_xhtml"><div class="ttname"><a href="classarmnn_1_1_i_connectable_layer.xhtml">armnn::IConnectableLayer</a></div><div class="ttdoc">Interface for a layer that is connectable to other layers via InputSlots and OutputSlots. </div><div class="ttdef"><b>Definition:</b> <a href="_i_network_8hpp_source.xhtml#l00068">INetwork.hpp:68</a></div></div> +<div class="ttc" id="classarmnn_1_1_profiler_manager_xhtml_a93857080c2523bf3395e7aa7e6024d5c"><div class="ttname"><a href="classarmnn_1_1_profiler_manager.xhtml#a93857080c2523bf3395e7aa7e6024d5c">armnn::ProfilerManager::GetInstance</a></div><div class="ttdeci">static ProfilerManager & GetInstance()</div><div class="ttdef"><b>Definition:</b> <a href="_profiling_8cpp_source.xhtml#l00572">Profiling.cpp:572</a></div></div> +<div class="ttc" id="classarmnn_1_1_tensor_info_xhtml"><div class="ttname"><a href="classarmnn_1_1_tensor_info.xhtml">armnn::TensorInfo</a></div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00152">Tensor.hpp:152</a></div></div> +<div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml">armnn::Convolution2dDescriptor</a></div><div class="ttdoc">A Convolution2dDescriptor for the Convolution2dLayer. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00495">Descriptors.hpp:495</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml_a1aa567f46c30960851c02847dc7b4215"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml#a1aa567f46c30960851c02847dc7b4215">armnn::NetworkImpl::AddConstantLayer</a></div><div class="ttdeci">IConnectableLayer * AddConstantLayer(const ConstTensor &input, const char *name=nullptr)</div><div class="ttdef"><b>Definition:</b> <a href="_network_8cpp_source.xhtml#l02285">Network.cpp:2285</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml_ab71c4df72f6587aea671acaacd6a0d9a"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml#ab71c4df72f6587aea671acaacd6a0d9a">armnn::NetworkImpl::AddConvertFp32ToFp16Layer</a></div><div class="ttdeci">IConnectableLayer * AddConvertFp32ToFp16Layer(const char *name=nullptr)</div><div class="ttdef"><b>Definition:</b> <a href="_network_8cpp_source.xhtml#l02084">Network.cpp:2084</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a150468a02bd7b2d2d061c4aaaee939f0"><div class="ttname"><a href="namespacearmnn.xhtml#a150468a02bd7b2d2d061c4aaaee939f0">armnn::IRuntimePtr</a></div><div class="ttdeci">std::unique_ptr< IRuntime, void(*)(IRuntime *runtime)> IRuntimePtr</div><div class="ttdef"><b>Definition:</b> <a href="_i_runtime_8hpp_source.xhtml#l00033">IRuntime.hpp:33</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_profiler_xhtml_a038bb767bbc6abc0ee0d9b509616b896"><div class="ttname"><a href="classarmnn_1_1_i_profiler.xhtml#a038bb767bbc6abc0ee0d9b509616b896">armnn::IProfiler::Print</a></div><div class="ttdeci">void Print(std::ostream &outStream) const</div><div class="ttdoc">Print stats for events in JSON Format to the given output stream. </div><div class="ttdef"><b>Definition:</b> <a href="_profiling_8cpp_source.xhtml#l00609">Profiling.cpp:609</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml_a383e74ef080d4a81b8b371be4b840248"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml#a383e74ef080d4a81b8b371be4b840248">armnn::NetworkImpl::AddConvolution2dLayer</a></div><div class="ttdeci">IConnectableLayer * AddConvolution2dLayer(const Convolution2dDescriptor &convolution2dDescriptor, const char *name=nullptr)</div><div class="ttdef"><b>Definition:</b> <a href="_network_8cpp_source.xhtml#l02047">Network.cpp:2047</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_aa01bce88f89975a5a031db4cc8861527"><div class="ttname"><a href="namespacearmnn.xhtml#aa01bce88f89975a5a031db4cc8861527">armnn::InputTensors</a></div><div class="ttdeci">std::vector< std::pair< LayerBindingId, class ConstTensor > > InputTensors</div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00392">Tensor.hpp:392</a></div></div> +<div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml_a002bc30e590d78cbb4f4d12171055ca7"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml#a002bc30e590d78cbb4f4d12171055ca7">armnn::Convolution2dDescriptor::m_PadRight</a></div><div class="ttdeci">uint32_t m_PadRight</div><div class="ttdoc">Padding right value in the width dimension. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00529">Descriptors.hpp:529</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_profiler_xhtml_ac9f22844fb2e329ffd193f2d9a8ce336"><div class="ttname"><a href="classarmnn_1_1_i_profiler.xhtml#ac9f22844fb2e329ffd193f2d9a8ce336">armnn::IProfiler::AnalyzeEventsAndWriteResults</a></div><div class="ttdeci">void AnalyzeEventsAndWriteResults(std::ostream &outStream) const</div><div class="ttdoc">Analyzes the tracked events and writes the results to the given output stream. </div><div class="ttdef"><b>Definition:</b> <a href="_profiling_8cpp_source.xhtml#l00604">Profiling.cpp:604</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml">armnn::NetworkImpl</a></div><div class="ttdoc">Private implementation of INetwork. </div><div class="ttdef"><b>Definition:</b> <a href="_network_8hpp_source.xhtml#l00031">Network.hpp:31</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_output_slot_xhtml_a5ee4a6c9a2481245487b1b1a70d20fd0"><div class="ttname"><a href="classarmnn_1_1_i_output_slot.xhtml#a5ee4a6c9a2481245487b1b1a70d20fd0">armnn::IOutputSlot::SetTensorInfo</a></div><div class="ttdeci">virtual void SetTensorInfo(const TensorInfo &tensorInfo)=0</div></div> +<div class="ttc" id="structarmnn_1_1_i_network_properties_xhtml"><div class="ttname"><a href="structarmnn_1_1_i_network_properties.xhtml">armnn::INetworkProperties</a></div><div class="ttdef"><b>Definition:</b> <a href="_i_runtime_8hpp_source.xhtml#l00035">IRuntime.hpp:35</a></div></div> +<div class="ttc" id="classarmnn_1_1_profiler_manager_xhtml_a3756986bc88b9b212d3f983c70c5c129"><div class="ttname"><a href="classarmnn_1_1_profiler_manager.xhtml#a3756986bc88b9b212d3f983c70c5c129">armnn::ProfilerManager::GetProfiler</a></div><div class="ttdeci">IProfiler * GetProfiler()</div><div class="ttdef"><b>Definition:</b> <a href="_profiling_8cpp_source.xhtml#l00584">Profiling.cpp:584</a></div></div> +<div class="ttc" id="classarmnn_1_1_tensor_xhtml"><div class="ttname"><a href="classarmnn_1_1_tensor.xhtml">armnn::Tensor</a></div><div class="ttdoc">A tensor defined by a TensorInfo (shape and data type) and a mutable backing store. </div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00319">Tensor.hpp:319</a></div></div> +<div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml_a56b51f56cef50cdfa554258eecdab046"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml#a56b51f56cef50cdfa554258eecdab046">armnn::Convolution2dDescriptor::m_PadTop</a></div><div class="ttdeci">uint32_t m_PadTop</div><div class="ttdoc">Padding top value in the height dimension. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00531">Descriptors.hpp:531</a></div></div> +<div class="ttc" id="_mem_copy_tests_8cpp_xhtml_a3df1acc0ccc35bce0bd6c027e23e2c45"><div class="ttname"><a href="_mem_copy_tests_8cpp.xhtml#a3df1acc0ccc35bce0bd6c027e23e2c45">TEST_CASE_FIXTURE</a></div><div class="ttdeci">TEST_CASE_FIXTURE(ClContextControlFixture, "CopyBetweenNeonAndGpu")</div><div class="ttdef"><b>Definition:</b> <a href="_mem_copy_tests_8cpp_source.xhtml#l00089">MemCopyTests.cpp:89</a></div></div> +<div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml_afe6a3377c4531315354def9023c8fdda"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml#afe6a3377c4531315354def9023c8fdda">armnn::Convolution2dDescriptor::m_StrideX</a></div><div class="ttdeci">uint32_t m_StrideX</div><div class="ttdoc">Stride value when proceeding through input for the width dimension. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00535">Descriptors.hpp:535</a></div></div> +<div class="ttc" id="structarmnn_1_1_optimizer_options_xhtml_a0054f53e4e70bb39c000bcf240627b18"><div class="ttname"><a href="structarmnn_1_1_optimizer_options.xhtml#a0054f53e4e70bb39c000bcf240627b18">armnn::OptimizerOptions::m_ExportEnabled</a></div><div class="ttdeci">bool m_ExportEnabled</div><div class="ttdef"><b>Definition:</b> <a href="_i_network_8hpp_source.xhtml#l00233">INetwork.hpp:233</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml_aa6c1c42ea44777302e87ce0fad5ac510"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml#aa6c1c42ea44777302e87ce0fad5ac510">armnn::NetworkImpl::AddInputLayer</a></div><div class="ttdeci">IConnectableLayer * AddInputLayer(LayerBindingId id, const char *name=nullptr)</div><div class="ttdef"><b>Definition:</b> <a href="_network_8cpp_source.xhtml#l01920">Network.cpp:1920</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a82e98ef05fd67036d1195ba17174d685"><div class="ttname"><a href="namespacearmnn.xhtml#a82e98ef05fd67036d1195ba17174d685">armnn::Optimize</a></div><div class="ttdeci">IOptimizedNetworkPtr Optimize(const INetwork &network, const std::vector< BackendId > &backendPreferences, const IDeviceSpec &deviceSpec, const OptimizerOptions &options=OptimizerOptions(), Optional< std::vector< std::string > &> messages=EmptyOptional())</div><div class="ttdoc">Create an optimized version of the network. </div><div class="ttdef"><b>Definition:</b> <a href="_network_8cpp_source.xhtml#l01864">Network.cpp:1864</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml_a2090bf6670b73c69309ed592068aa3af"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml#a2090bf6670b73c69309ed592068aa3af">armnn::NetworkImpl::AddConvertFp16ToFp32Layer</a></div><div class="ttdeci">IConnectableLayer * AddConvertFp16ToFp32Layer(const char *name=nullptr)</div><div class="ttdef"><b>Definition:</b> <a href="_network_8cpp_source.xhtml#l02079">Network.cpp:2079</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_ae2f04a162585c0a5222a537efd5456aeaec0fc0100c4fc1ce4eea230c3dc10360"><div class="ttname"><a href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeaec0fc0100c4fc1ce4eea230c3dc10360">armnn::Compute::Undefined</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a0d8160388a127c1a23b37bc88dc6e2ec"><div class="ttname"><a href="namespacearmnn.xhtml#a0d8160388a127c1a23b37bc88dc6e2ec">armnn::NetworkId</a></div><div class="ttdeci">int NetworkId</div><div class="ttdef"><b>Definition:</b> <a href="_i_runtime_8hpp_source.xhtml#l00027">IRuntime.hpp:27</a></div></div> +<div class="ttc" id="classarmnn_1_1_const_tensor_xhtml"><div class="ttname"><a href="classarmnn_1_1_const_tensor.xhtml">armnn::ConstTensor</a></div><div class="ttdoc">A tensor defined by a TensorInfo (shape and data type) and an immutable backing store. </div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00327">Tensor.hpp:327</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a8f091a512915d1cb29a4ebf13dfc53ea"><div class="ttname"><a href="namespacearmnn.xhtml#a8f091a512915d1cb29a4ebf13dfc53ea">armnn::OutputTensors</a></div><div class="ttdeci">std::vector< std::pair< LayerBindingId, class Tensor > > OutputTensors</div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00393">Tensor.hpp:393</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_ad8ed01ff3ff33333d8e19db4d2818bb6a26e6ed77470c6f2f830ecf874e6c0d55"><div class="ttname"><a href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a26e6ed77470c6f2f830ecf874e6c0d55">armnn::DataType::Float16</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a674efcf6cbdb9e831d653ff0e821fb38"><div class="ttname"><a href="namespacearmnn.xhtml#a674efcf6cbdb9e831d653ff0e821fb38">armnn::IOptimizedNetworkPtr</a></div><div class="ttdeci">std::unique_ptr< IOptimizedNetwork, void(*)(IOptimizedNetwork *network)> IOptimizedNetworkPtr</div><div class="ttdef"><b>Definition:</b> <a href="_i_network_8hpp_source.xhtml#l00239">INetwork.hpp:239</a></div></div> +<div class="ttc" id="classarmnn_1_1_profiler_manager_xhtml"><div class="ttname"><a href="classarmnn_1_1_profiler_manager.xhtml">armnn::ProfilerManager</a></div><div class="ttdef"><b>Definition:</b> <a href="_profiling_8hpp_source.xhtml#l00111">Profiling.hpp:111</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml_af5790069aa11fd1c5bb2e17cecb06528"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml#af5790069aa11fd1c5bb2e17cecb06528">armnn::NetworkImpl::AddOutputLayer</a></div><div class="ttdeci">IConnectableLayer * AddOutputLayer(LayerBindingId id, const char *name=nullptr)</div><div class="ttdef"><b>Definition:</b> <a href="_network_8cpp_source.xhtml#l02224">Network.cpp:2224</a></div></div> +<div class="ttc" id="_assert_8hpp_xhtml_a5698be69cbd5dfe6c28fcd9867e8cbed"><div class="ttname"><a href="_assert_8hpp.xhtml#a5698be69cbd5dfe6c28fcd9867e8cbed">ARMNN_ASSERT</a></div><div class="ttdeci">#define ARMNN_ASSERT(COND)</div><div class="ttdef"><b>Definition:</b> <a href="_assert_8hpp_source.xhtml#l00014">Assert.hpp:14</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a14fcd7f88d11cea0a018269dca5f9277a1131a914388fac73e5f07b0ba0aad523"><div class="ttname"><a href="namespacearmnn.xhtml#a14fcd7f88d11cea0a018269dca5f9277a1131a914388fac73e5f07b0ba0aad523">armnn::MemorySource::Malloc</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b"><div class="ttname"><a href="namespacearmnn.xhtml#ae2f04a162585c0a5222a537efd5456aeafaa4524e3df19ada32643ce9a222362b">armnn::Compute::GpuAcc</a></div><div class="ttdoc">GPU Execution: OpenCL: ArmCompute. </div></div> +<div class="ttc" id="structarmnn_1_1_optimizer_options_xhtml"><div class="ttname"><a href="structarmnn_1_1_optimizer_options.xhtml">armnn::OptimizerOptions</a></div><div class="ttdoc">ArmNN performs an optimization on each model/network before it gets loaded for execution. </div><div class="ttdef"><b>Definition:</b> <a href="_i_network_8hpp_source.xhtml#l00127">INetwork.hpp:127</a></div></div> +<div class="ttc" id="structarmnn_1_1_activation_descriptor_xhtml"><div class="ttname"><a href="structarmnn_1_1_activation_descriptor.xhtml">armnn::ActivationDescriptor</a></div><div class="ttdoc">An ActivationDescriptor for the ActivationLayer. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00036">Descriptors.hpp:36</a></div></div> +<div class="ttc" id="classarmnn_1_1_cl_import_tensor_handle_factory_xhtml"><div class="ttname"><a href="classarmnn_1_1_cl_import_tensor_handle_factory.xhtml">armnn::ClImportTensorHandleFactory</a></div><div class="ttdoc">This factory creates ClImportTensorHandles that refer to imported memory tensors. ...</div><div class="ttdef"><b>Definition:</b> <a href="_cl_import_tensor_handle_factory_8hpp_source.xhtml#l00023">ClImportTensorHandleFactory.hpp:23</a></div></div> +<div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml_ac1fe174bbadfb39a2b636940c2e647c8"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac1fe174bbadfb39a2b636940c2e647c8">armnn::Convolution2dDescriptor::m_StrideY</a></div><div class="ttdeci">uint32_t m_StrideY</div><div class="ttdoc">Stride value when proceeding through input for the height dimension. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00537">Descriptors.hpp:537</a></div></div> +<div class="ttc" id="structarmnn_1_1_optimizer_options_xhtml_a05c1bba6ba3ecc1339d4c4c10c0d8890"><div class="ttname"><a href="structarmnn_1_1_optimizer_options.xhtml#a05c1bba6ba3ecc1339d4c4c10c0d8890">armnn::OptimizerOptions::m_ImportEnabled</a></div><div class="ttdeci">bool m_ImportEnabled</div><div class="ttdef"><b>Definition:</b> <a href="_i_network_8hpp_source.xhtml#l00224">INetwork.hpp:224</a></div></div> +<div class="ttc" id="structarmnn_1_1_i_runtime_1_1_creation_options_xhtml"><div class="ttname"><a href="structarmnn_1_1_i_runtime_1_1_creation_options.xhtml">armnn::IRuntime::CreationOptions</a></div><div class="ttdef"><b>Definition:</b> <a href="_i_runtime_8hpp_source.xhtml#l00077">IRuntime.hpp:77</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c"><div class="ttname"><a href="namespacearmnn.xhtml#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">armnn::BoostLogSeverityMapping::info</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_a14fcd7f88d11cea0a018269dca5f9277"><div class="ttname"><a href="namespacearmnn.xhtml#a14fcd7f88d11cea0a018269dca5f9277">armnn::MemorySource</a></div><div class="ttdeci">MemorySource</div><div class="ttdoc">Define the Memory Source to reduce copies. </div><div class="ttdef"><b>Definition:</b> <a href="_types_8hpp_source.xhtml#l00230">Types.hpp:230</a></div></div> +<div class="ttc" id="classarmnn_1_1_memory_import_exception_xhtml"><div class="ttname"><a href="classarmnn_1_1_memory_import_exception.xhtml">armnn::MemoryImportException</a></div><div class="ttdef"><b>Definition:</b> <a href="_exceptions_8hpp_source.xhtml#l00125">Exceptions.hpp:125</a></div></div> +<div class="ttc" id="classarmnn_1_1_network_impl_xhtml_afe0a4f719f9752a405e71878da7012ba"><div class="ttname"><a href="classarmnn_1_1_network_impl.xhtml#afe0a4f719f9752a405e71878da7012ba">armnn::NetworkImpl::GetGraph</a></div><div class="ttdeci">const Graph & GetGraph() const</div><div class="ttdef"><b>Definition:</b> <a href="_network_8hpp_source.xhtml#l00037">Network.hpp:37</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_connectable_layer_xhtml_a6ec9e0eb66d7d6a01240492a0b18104c"><div class="ttname"><a href="classarmnn_1_1_i_connectable_layer.xhtml#a6ec9e0eb66d7d6a01240492a0b18104c">armnn::IConnectableLayer::GetInputSlot</a></div><div class="ttdeci">virtual const IInputSlot & GetInputSlot(unsigned int index) const =0</div><div class="ttdoc">Get a const input slot handle by slot index. </div></div> +<div class="ttc" id="classarmnn_1_1_tensor_info_xhtml_a8ffca1e21bdfa7f945617acd606aac91"><div class="ttname"><a href="classarmnn_1_1_tensor_info.xhtml#a8ffca1e21bdfa7f945617acd606aac91">armnn::TensorInfo::SetConstant</a></div><div class="ttdeci">void SetConstant(const bool IsConstant=true)</div><div class="ttdoc">Marks the data corresponding to this tensor info as constant. </div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8cpp_source.xhtml#l00514">Tensor.cpp:514</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_ad8ed01ff3ff33333d8e19db4d2818bb6a166495adc0d0f53bee6baecc577f5204"><div class="ttname"><a href="namespacearmnn.xhtml#ad8ed01ff3ff33333d8e19db4d2818bb6a166495adc0d0f53bee6baecc577f5204">armnn::DataType::Float32</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_connectable_layer_xhtml_a80ac4eda2e7f2757ec9dd96fc96dbd16"><div class="ttname"><a href="classarmnn_1_1_i_connectable_layer.xhtml#a80ac4eda2e7f2757ec9dd96fc96dbd16">armnn::IConnectableLayer::GetOutputSlot</a></div><div class="ttdeci">virtual const IOutputSlot & GetOutputSlot(unsigned int index) const =0</div><div class="ttdoc">Get the const output slot handle by slot index. </div></div> +<div class="ttc" id="struct_cl_context_control_fixture_base_xhtml"><div class="ttname"><a href="struct_cl_context_control_fixture_base.xhtml">ClContextControlFixtureBase</a></div><div class="ttdef"><b>Definition:</b> <a href="_cl_context_control_fixture_8hpp_source.xhtml#l00012">ClContextControlFixture.hpp:12</a></div></div> +<div class="ttc" id="namespacearmnn_xhtml_ace74f6f9feb95a964a49d79458232703"><div class="ttname"><a href="namespacearmnn.xhtml#ace74f6f9feb95a964a49d79458232703">armnn::INetworkPtr</a></div><div class="ttdeci">std::unique_ptr< INetwork, void(*)(INetwork *network)> INetworkPtr</div><div class="ttdef"><b>Definition:</b> <a href="_i_network_8hpp_source.xhtml#l00238">INetwork.hpp:238</a></div></div> +<div class="ttc" id="classarmnn_1_1_i_output_slot_xhtml_ac1835f8756a9f03c02fcf9664e3a0fce"><div class="ttname"><a href="classarmnn_1_1_i_output_slot.xhtml#ac1835f8756a9f03c02fcf9664e3a0fce">armnn::IOutputSlot::Connect</a></div><div class="ttdeci">virtual int Connect(IInputSlot &destination)=0</div></div> +<div class="ttc" id="namespacearmnn_xhtml_a0f38fa92b2468d5378258a2b074c1a31"><div class="ttname"><a href="namespacearmnn.xhtml#a0f38fa92b2468d5378258a2b074c1a31">armnn::Half</a></div><div class="ttdeci">half_float::half Half</div><div class="ttdef"><b>Definition:</b> <a href="_half_8hpp_source.xhtml#l00018">Half.hpp:18</a></div></div> +<div class="ttc" id="structarmnn_1_1_activation_descriptor_xhtml_af10fa7883e3579950f477bee92a64844"><div class="ttname"><a href="structarmnn_1_1_activation_descriptor.xhtml#af10fa7883e3579950f477bee92a64844">armnn::ActivationDescriptor::m_Function</a></div><div class="ttdeci">ActivationFunction m_Function</div><div class="ttdoc">The activation function to use (Sigmoid, TanH, Linear, ReLu, BoundedReLu, SoftReLu, LeakyReLu, Abs, Sqrt, Square, Elu). </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00059">Descriptors.hpp:59</a></div></div> +<div class="ttc" id="structarmnn_1_1_convolution2d_descriptor_xhtml_ac18546ebbebbb32fe0a03baa9bf2c600"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.xhtml#ac18546ebbebbb32fe0a03baa9bf2c600">armnn::Convolution2dDescriptor::m_PadLeft</a></div><div class="ttdeci">uint32_t m_PadLeft</div><div class="ttdoc">Padding left value in the width dimension. </div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.xhtml#l00527">Descriptors.hpp:527</a></div></div> +<div class="ttc" id="classarmnn_1_1_tensor_info_xhtml_a8846406ac37fbd2204f0be16ee05d5b7"><div class="ttname"><a href="classarmnn_1_1_tensor_info.xhtml#a8846406ac37fbd2204f0be16ee05d5b7">armnn::TensorInfo::GetNumElements</a></div><div class="ttdeci">unsigned int GetNumElements() const</div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.xhtml#l00196">Tensor.hpp:196</a></div></div> +</div><!-- fragment --> +</div> +</div> +</div><!-- contents --> +</div><!-- doc-content --> +<!-- start footer part --> +<div id="nav-path" class="navpath"><!-- id is needed for treeview function! --> + <ul> + <li class="navelem"><a class="el" href="dir_68267d1309a1af8e8297ef4c3efbcdba.xhtml">src</a></li><li class="navelem"><a class="el" href="dir_0f3cdec46afbc61a1ded8e1687c9c9a0.xhtml">backends</a></li><li class="navelem"><a class="el" href="dir_1ad86c6d39ab715a831555571b9e98a5.xhtml">cl</a></li><li class="navelem"><a class="el" href="dir_02bab2737bbb2fb3882a0be567244fbf.xhtml">test</a></li><li class="navelem"><a class="el" href="_cl_import_tensor_handle_tests_8cpp.xhtml">ClImportTensorHandleTests.cpp</a></li> + <li class="footer">Generated on Fri Aug 19 2022 14:38:33 for ArmNN by + <a href="http://www.doxygen.org/index.html"> + <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.13 </li> + </ul> +</div> +</body> +</html> |