Nikhil Raj | 1dc83fe | 2024-05-16 09:47:51 +0100 | [diff] [blame^] | 1 | <!-- HTML header for doxygen 1.8.17--> |
| 2 | <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "https://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> |
| 3 | <html xmlns="http://www.w3.org/1999/xhtml"> |
| 4 | <head> |
| 5 | <meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/> |
| 6 | <meta http-equiv="X-UA-Compatible" content="IE=9"/> |
| 7 | <meta name="generator" content="Doxygen 1.8.17"/> |
| 8 | <meta name="viewport" content="width=device-width, initial-scale=1"/> |
| 9 | <title>Arm NN: src/backends/gpuFsa/workloads/GpuFsaPreCompiledWorkload.cpp Source File</title> |
| 10 | <link href="tabs.css" rel="stylesheet" type="text/css"/> |
| 11 | <script type="text/javascript" src="jquery.js"></script> |
| 12 | <script type="text/javascript" src="dynsections.js"></script> |
| 13 | <link href="navtree.css" rel="stylesheet" type="text/css"/> |
| 14 | <script type="text/javascript" src="resize.js"></script> |
| 15 | <script type="text/javascript" src="navtreedata.js"></script> |
| 16 | <script type="text/javascript" src="navtree.js"></script> |
| 17 | <link href="search/search.css" rel="stylesheet" type="text/css"/> |
| 18 | <script type="text/javascript" src="search/searchdata.js"></script> |
| 19 | <script type="text/javascript" src="search/search.js"></script> |
| 20 | <script type="text/x-mathjax-config"> |
| 21 | MathJax.Hub.Config({ |
| 22 | extensions: ["tex2jax.js"], |
| 23 | jax: ["input/TeX","output/HTML-CSS"], |
| 24 | }); |
| 25 | </script> |
| 26 | <script type="text/javascript" async="async" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script> |
| 27 | <link href="doxygen.css" rel="stylesheet" type="text/css" /> |
| 28 | <link href="customdoxygen.css" rel="stylesheet" type="text/css"/> |
| 29 | </head> |
| 30 | <body> |
| 31 | <div id="top"><!-- do not remove this div, it is closed by doxygen! --> |
| 32 | <div id="titlearea"> |
| 33 | <table cellspacing="0" cellpadding="0"> |
| 34 | <tbody> |
| 35 | <tr style="height: 56px;"> |
| 36 | <img alt="ArmNN" src="Arm_NN_horizontal_blue.png" style="max-width: 15rem; margin-top: .5rem; margin-left 13px"/> |
| 37 | <td id="projectalign" style="padding-left: 0.9em;"> |
| 38 | <div id="projectname"> |
| 39 |  <span id="projectnumber">24.05</span> |
| 40 | </div> |
| 41 | </td> |
| 42 | </tr> |
| 43 | </tbody> |
| 44 | </table> |
| 45 | </div> |
| 46 | <!-- end header part --> |
| 47 | <!-- Generated by Doxygen 1.8.17 --> |
| 48 | <script type="text/javascript"> |
| 49 | /* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */ |
| 50 | var searchBox = new SearchBox("searchBox", "search",false,'Search'); |
| 51 | /* @license-end */ |
| 52 | </script> |
| 53 | <script type="text/javascript" src="menudata.js"></script> |
| 54 | <script type="text/javascript" src="menu.js"></script> |
| 55 | <script type="text/javascript"> |
| 56 | /* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */ |
| 57 | $(function() { |
| 58 | initMenu('',true,false,'search.php','Search'); |
| 59 | $(document).ready(function() { init_search(); }); |
| 60 | }); |
| 61 | /* @license-end */</script> |
| 62 | <div id="main-nav"></div> |
| 63 | </div><!-- top --> |
| 64 | <div id="side-nav" class="ui-resizable side-nav-resizable"> |
| 65 | <div id="nav-tree"> |
| 66 | <div id="nav-tree-contents"> |
| 67 | <div id="nav-sync" class="sync"></div> |
| 68 | </div> |
| 69 | </div> |
| 70 | <div id="splitbar" style="-moz-user-select:none;" |
| 71 | class="ui-resizable-handle"> |
| 72 | </div> |
| 73 | </div> |
| 74 | <script type="text/javascript"> |
| 75 | /* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */ |
| 76 | $(document).ready(function(){initNavTree('_gpu_fsa_pre_compiled_workload_8cpp_source.html',''); initResizable(); }); |
| 77 | /* @license-end */ |
| 78 | </script> |
| 79 | <div id="doc-content"> |
| 80 | <!-- window showing the filter options --> |
| 81 | <div id="MSearchSelectWindow" |
| 82 | onmouseover="return searchBox.OnSearchSelectShow()" |
| 83 | onmouseout="return searchBox.OnSearchSelectHide()" |
| 84 | onkeydown="return searchBox.OnSearchSelectKey(event)"> |
| 85 | </div> |
| 86 | |
| 87 | <!-- iframe showing the search results (closed by default) --> |
| 88 | <div id="MSearchResultsWindow"> |
| 89 | <iframe src="javascript:void(0)" frameborder="0" |
| 90 | name="MSearchResults" id="MSearchResults"> |
| 91 | </iframe> |
| 92 | </div> |
| 93 | |
| 94 | <div class="header"> |
| 95 | <div class="headertitle"> |
| 96 | <div class="title">GpuFsaPreCompiledWorkload.cpp</div> </div> |
| 97 | </div><!--header--> |
| 98 | <div class="contents"> |
| 99 | <a href="_gpu_fsa_pre_compiled_workload_8cpp.html">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> <span class="comment">//</span></div> |
| 100 | <div class="line"><a name="l00002"></a><span class="lineno"> 2</span> <span class="comment">// Copyright © 2024 Arm Ltd and Contributors. All rights reserved.</span></div> |
| 101 | <div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment">// SPDX-License-Identifier: MIT</span></div> |
| 102 | <div class="line"><a name="l00004"></a><span class="lineno"> 4</span> <span class="comment">//</span></div> |
| 103 | <div class="line"><a name="l00005"></a><span class="lineno"> 5</span>  </div> |
| 104 | <div class="line"><a name="l00006"></a><span class="lineno"> 6</span> <span class="preprocessor">#include "<a class="code" href="_gpu_fsa_pre_compiled_workload_8hpp.html">GpuFsaPreCompiledWorkload.hpp</a>"</span></div> |
| 105 | <div class="line"><a name="l00007"></a><span class="lineno"> 7</span> <span class="preprocessor">#include "<a class="code" href="_gpu_fsa_workload_utils_8hpp.html">GpuFsaWorkloadUtils.hpp</a>"</span></div> |
| 106 | <div class="line"><a name="l00008"></a><span class="lineno"> 8</span> <span class="preprocessor">#include "<a class="code" href="_polymorphic_downcast_8hpp.html">armnn/utility/PolymorphicDowncast.hpp</a>"</span></div> |
| 107 | <div class="line"><a name="l00009"></a><span class="lineno"> 9</span>  </div> |
| 108 | <div class="line"><a name="l00010"></a><span class="lineno"> 10</span> <span class="preprocessor">#include <<a class="code" href="_gpu_fsa_tensor_handle_8hpp.html">gpuFsa/GpuFsaTensorHandle.hpp</a>></span></div> |
| 109 | <div class="line"><a name="l00011"></a><span class="lineno"> 11</span> <span class="preprocessor">#include <<a class="code" href="_gpu_fsa_backend_8hpp.html">gpuFsa/GpuFsaBackend.hpp</a>></span></div> |
| 110 | <div class="line"><a name="l00012"></a><span class="lineno"> 12</span> <span class="preprocessor">#include <<a class="code" href="_arm_compute_tensor_utils_8hpp.html">aclCommon/ArmComputeTensorUtils.hpp</a>></span></div> |
| 111 | <div class="line"><a name="l00013"></a><span class="lineno"> 13</span> <span class="preprocessor">#include <fmt/format.h></span></div> |
| 112 | <div class="line"><a name="l00014"></a><span class="lineno"> 14</span>  </div> |
| 113 | <div class="line"><a name="l00015"></a><span class="lineno"> 15</span> <span class="preprocessor">#include <<a class="code" href="_arm_compute_tensor_utils_8hpp.html">aclCommon/ArmComputeTensorUtils.hpp</a>></span></div> |
| 114 | <div class="line"><a name="l00016"></a><span class="lineno"> 16</span> <span class="preprocessor">#include <arm_compute/runtime/CL/CLTensor.h></span></div> |
| 115 | <div class="line"><a name="l00017"></a><span class="lineno"> 17</span> <span class="preprocessor">#include <arm_compute/core/ITensorInfo.h></span></div> |
| 116 | <div class="line"><a name="l00018"></a><span class="lineno"> 18</span> <span class="preprocessor">#include <arm_compute/core/TensorInfo.h></span></div> |
| 117 | <div class="line"><a name="l00019"></a><span class="lineno"> 19</span> <span class="preprocessor">#include <arm_compute/core/TensorShape.h></span></div> |
| 118 | <div class="line"><a name="l00020"></a><span class="lineno"> 20</span> <span class="preprocessor">#include <arm_compute/core/CL/CLKernelLibrary.h></span></div> |
| 119 | <div class="line"><a name="l00021"></a><span class="lineno"> 21</span> <span class="preprocessor">#include <arm_compute/core/CL/CLCompileContext.h></span></div> |
| 120 | <div class="line"><a name="l00022"></a><span class="lineno"> 22</span>  </div> |
| 121 | <div class="line"><a name="l00023"></a><span class="lineno"> 23</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/runtime/gpu/cl/ClWorkloadRuntime.h></span></div> |
| 122 | <div class="line"><a name="l00024"></a><span class="lineno"> 24</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/GpuWorkloadContext.h></span></div> |
| 123 | <div class="line"><a name="l00025"></a><span class="lineno"> 25</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/operators/GpuConv2d.h></span></div> |
| 124 | <div class="line"><a name="l00026"></a><span class="lineno"> 26</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/operators/GpuOutput.h></span></div> |
| 125 | <div class="line"><a name="l00027"></a><span class="lineno"> 27</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/GpuWorkloadSketch.h></span></div> |
| 126 | <div class="line"><a name="l00028"></a><span class="lineno"> 28</span>  </div> |
| 127 | <div class="line"><a name="l00029"></a><span class="lineno"> 29</span> <span class="keyword">namespace </span><a class="code" href="namespacearmnn.html">armnn</a> {</div> |
| 128 | <div class="line"><a name="l00030"></a><span class="lineno"> 30</span>  </div> |
| 129 | <div class="line"><a name="l00031"></a><span class="lineno"><a class="line" href="classarmnn_1_1_gpu_fsa_pre_compiled_workload.html#aeef6e7488b495adc96f8abd7d29e3d5f"> 31</a></span> <a class="code" href="classarmnn_1_1_gpu_fsa_pre_compiled_workload.html#aeef6e7488b495adc96f8abd7d29e3d5f">GpuFsaPreCompiledWorkload::GpuFsaPreCompiledWorkload</a>(<span class="keyword">const</span> <a class="code" href="structarmnn_1_1_pre_compiled_queue_descriptor.html">PreCompiledQueueDescriptor</a> &descriptor,</div> |
| 130 | <div class="line"><a name="l00032"></a><span class="lineno"> 32</span>  <span class="keyword">const</span> <a class="code" href="structarmnn_1_1_workload_info.html">WorkloadInfo</a> &info)</div> |
| 131 | <div class="line"><a name="l00033"></a><span class="lineno"> 33</span>  : <a class="code" href="classarmnn_1_1_base_workload.html">BaseWorkload</a><<a class="code" href="structarmnn_1_1_pre_compiled_queue_descriptor.html">PreCompiledQueueDescriptor</a>>(descriptor, <a class="code" href="namespacearmnn.html#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>), m_workloadInfo(<a class="code" href="namespacearmnn.html#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>)</div> |
| 132 | <div class="line"><a name="l00034"></a><span class="lineno"> 34</span> {</div> |
| 133 | <div class="line"><a name="l00035"></a><span class="lineno"> 35</span>  <span class="comment">// Check that the workload is holding a pointer to a valid pre-compiled object</span></div> |
| 134 | <div class="line"><a name="l00036"></a><span class="lineno"> 36</span>  <span class="keywordflow">if</span> (<a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_pre_compiled_queue_descriptor.html#aa1fedd1a6371526cb68cc5cc58c87465">m_PreCompiledObject</a> == <span class="keyword">nullptr</span>)</div> |
| 135 | <div class="line"><a name="l00037"></a><span class="lineno"> 37</span>  {</div> |
| 136 | <div class="line"><a name="l00038"></a><span class="lineno"> 38</span>  <span class="keywordflow">throw</span> <a class="code" href="classarmnn_1_1_invalid_argument_exception.html">InvalidArgumentException</a>(</div> |
| 137 | <div class="line"><a name="l00039"></a><span class="lineno"> 39</span>  <span class="stringliteral">"GpuFsaPrecompiledWorkload requires a valid pre-compiled object (GpuWorkloadSketch)."</span>);</div> |
| 138 | <div class="line"><a name="l00040"></a><span class="lineno"> 40</span>  }</div> |
| 139 | <div class="line"><a name="l00041"></a><span class="lineno"> 41</span> }</div> |
| 140 | <div class="line"><a name="l00042"></a><span class="lineno"> 42</span>  </div> |
| 141 | <div class="line"><a name="l00043"></a><span class="lineno"><a class="line" href="classarmnn_1_1_gpu_fsa_pre_compiled_workload.html#ae071e8822437c78baea75c3aef3a263a"> 43</a></span> <span class="keywordtype">void</span> <a class="code" href="classarmnn_1_1_gpu_fsa_pre_compiled_workload.html#ae071e8822437c78baea75c3aef3a263a">GpuFsaPreCompiledWorkload::Execute</a>()<span class="keyword"> const</span></div> |
| 142 | <div class="line"><a name="l00044"></a><span class="lineno"> 44</span> <span class="keyword"></span>{</div> |
| 143 | <div class="line"><a name="l00045"></a><span class="lineno"> 45</span> <span class="comment">/*</span></div> |
| 144 | <div class="line"><a name="l00046"></a><span class="lineno"> 46</span> <span class="comment"> * The Execute function of the GpuFsa Backends PreCompiled workload needs to jump through various hoops in order to</span></div> |
| 145 | <div class="line"><a name="l00047"></a><span class="lineno"> 47</span> <span class="comment"> * create a valid sketch and runtime that can execute the kernel</span></div> |
| 146 | <div class="line"><a name="l00048"></a><span class="lineno"> 48</span> <span class="comment"> * First we need all of the data stored within the PreCompiled blob which was used to setup the workload, namely:</span></div> |
| 147 | <div class="line"><a name="l00049"></a><span class="lineno"> 49</span> <span class="comment"> * The GpuWorkloadContext, this is a context which contains the TensorInfos and is unique to the graph being run</span></div> |
| 148 | <div class="line"><a name="l00050"></a><span class="lineno"> 50</span> <span class="comment"> * The Sketch, this can contain one or many ops and acts as a subgraph within the context</span></div> |
| 149 | <div class="line"><a name="l00051"></a><span class="lineno"> 51</span> <span class="comment"> * The inputTensorInfos / outputTensorInfos, These are vectors containing the TensorInfos used when creating the sketch</span></div> |
| 150 | <div class="line"><a name="l00052"></a><span class="lineno"> 52</span> <span class="comment"> *</span></div> |
| 151 | <div class="line"><a name="l00053"></a><span class="lineno"> 53</span> <span class="comment"> * It is very important that the Tensors passed into the Runtime being used to execute this sketch are created with</span></div> |
| 152 | <div class="line"><a name="l00054"></a><span class="lineno"> 54</span> <span class="comment"> * the same TensorInfos as used when creating the sketch. We do this by creating new tensors, getting the original</span></div> |
| 153 | <div class="line"><a name="l00055"></a><span class="lineno"> 55</span> <span class="comment"> * TensorInfos from the vectors of tensorInfos stored in the blob, and then importing the buffers from our own</span></div> |
| 154 | <div class="line"><a name="l00056"></a><span class="lineno"> 56</span> <span class="comment"> * TensorHandles directly into these newly created Tensors. This allows us to link the externally visible Tensors</span></div> |
| 155 | <div class="line"><a name="l00057"></a><span class="lineno"> 57</span> <span class="comment"> * from ArmNN to the Tensors which are needed to execute with the Sketch.</span></div> |
| 156 | <div class="line"><a name="l00058"></a><span class="lineno"> 58</span> <span class="comment"> *</span></div> |
| 157 | <div class="line"><a name="l00059"></a><span class="lineno"> 59</span> <span class="comment"> */</span></div> |
| 158 | <div class="line"><a name="l00060"></a><span class="lineno"> 60</span>  <span class="keyword">using namespace </span>arm_compute::experimental::dynamic_fusion;</div> |
| 159 | <div class="line"><a name="l00061"></a><span class="lineno"> 61</span>  <span class="comment">// Get the runtime and configure it with the precompiled sketch</span></div> |
| 160 | <div class="line"><a name="l00062"></a><span class="lineno"> 62</span>  ClWorkloadRuntime runtime;</div> |
| 161 | <div class="line"><a name="l00063"></a><span class="lineno"> 63</span>  <a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html">GpuFsaPreCompiledBlob</a> *preCompiledBlob = <span class="keyword">static_cast<</span><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html">GpuFsaPreCompiledBlob</a>*<span class="keyword">></span>(<a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_pre_compiled_queue_descriptor.html#aa1fedd1a6371526cb68cc5cc58c87465">m_PreCompiledObject</a>);</div> |
| 162 | <div class="line"><a name="l00064"></a><span class="lineno"> 64</span>  <span class="keyword">auto</span> sketch = preCompiledBlob-><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a14f92a9f65e32c3da896e7b1d45abd02">sketch</a>.release();</div> |
| 163 | <div class="line"><a name="l00065"></a><span class="lineno"> 65</span>  <span class="keyword">auto</span> status = runtime.configure(*sketch);</div> |
| 164 | <div class="line"><a name="l00066"></a><span class="lineno"> 66</span>  </div> |
| 165 | <div class="line"><a name="l00067"></a><span class="lineno"> 67</span>  <span class="comment">// Get the TensorInfos stored within the PreCompiledBlob and check they're the right size</span></div> |
| 166 | <div class="line"><a name="l00068"></a><span class="lineno"> 68</span>  <span class="keyword">auto</span> inputTensorInfos = preCompiledBlob-><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a479b90f0b24c427502d94b716117e034">inputTensorInfos</a>.get();</div> |
| 167 | <div class="line"><a name="l00069"></a><span class="lineno"> 69</span>  <span class="keyword">auto</span> outputTensorInfos = preCompiledBlob-><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#ac49bf679a23aa84f06a6bde3440a4c40">outputTensorInfos</a>.get();</div> |
| 168 | <div class="line"><a name="l00070"></a><span class="lineno"> 70</span>  <span class="keywordflow">if</span> (inputTensorInfos->size() != <a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a4b50e46a6810018f3edecfb68b2a76b3">m_Inputs</a>.size())</div> |
| 169 | <div class="line"><a name="l00071"></a><span class="lineno"> 71</span>  {</div> |
| 170 | <div class="line"><a name="l00072"></a><span class="lineno"> 72</span>  <span class="keywordflow">throw</span> <a class="code" href="classarmnn_1_1_invalid_argument_exception.html">InvalidArgumentException</a>(fmt::format(<span class="stringliteral">"GpuFsaPreCompiledWorkload::Execute: The number of inputTensorInfos"</span></div> |
| 171 | <div class="line"><a name="l00073"></a><span class="lineno"> 73</span>  <span class="stringliteral">" {} does not match the number of inputs {}."</span>,</div> |
| 172 | <div class="line"><a name="l00074"></a><span class="lineno"> 74</span>  inputTensorInfos->size(), <a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a4b50e46a6810018f3edecfb68b2a76b3">m_Inputs</a>.size()));</div> |
| 173 | <div class="line"><a name="l00075"></a><span class="lineno"> 75</span>  }</div> |
| 174 | <div class="line"><a name="l00076"></a><span class="lineno"> 76</span>  <span class="keywordflow">if</span> (outputTensorInfos->size() != <a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a6abd491bb99ffe88bd472c1ae5a1ed1a">m_Outputs</a>.size())</div> |
| 175 | <div class="line"><a name="l00077"></a><span class="lineno"> 77</span>  {</div> |
| 176 | <div class="line"><a name="l00078"></a><span class="lineno"> 78</span>  <span class="keywordflow">throw</span> <a class="code" href="classarmnn_1_1_invalid_argument_exception.html">InvalidArgumentException</a>(fmt::format(<span class="stringliteral">"GpuFsaPreCompiledWorkload::Execute: The number of outputTensorInfos"</span></div> |
| 177 | <div class="line"><a name="l00079"></a><span class="lineno"> 79</span>  <span class="stringliteral">" {} does not match the number of outputs {}."</span>,</div> |
| 178 | <div class="line"><a name="l00080"></a><span class="lineno"> 80</span>  outputTensorInfos->size(), <a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a6abd491bb99ffe88bd472c1ae5a1ed1a">m_Outputs</a>.size()));</div> |
| 179 | <div class="line"><a name="l00081"></a><span class="lineno"> 81</span>  }</div> |
| 180 | <div class="line"><a name="l00082"></a><span class="lineno"> 82</span>  </div> |
| 181 | <div class="line"><a name="l00083"></a><span class="lineno"> 83</span>  <span class="comment">// (Important) Allocate auxiliary tensor memory if there are any</span></div> |
| 182 | <div class="line"><a name="l00084"></a><span class="lineno"> 84</span>  <span class="keywordflow">for</span>(<span class="keyword">auto</span> &data : runtime.get_auxiliary_tensors())</div> |
| 183 | <div class="line"><a name="l00085"></a><span class="lineno"> 85</span>  {</div> |
| 184 | <div class="line"><a name="l00086"></a><span class="lineno"> 86</span>  arm_compute::CLTensor* tensor = std::get<0>(data);</div> |
| 185 | <div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  arm_compute::TensorInfo <a class="code" href="namespacearmnn.html#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a> = std::get<1>(data);</div> |
| 186 | <div class="line"><a name="l00088"></a><span class="lineno"> 88</span>  arm_compute::experimental::dynamic_fusion::AuxMemoryInfo aux_mem_req = std::get<2>(data);</div> |
| 187 | <div class="line"><a name="l00089"></a><span class="lineno"> 89</span>  tensor->allocator()->init(<a class="code" href="namespacearmnn.html#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">info</a>, aux_mem_req.alignment);</div> |
| 188 | <div class="line"><a name="l00090"></a><span class="lineno"> 90</span>  tensor->allocator()->allocate(); <span class="comment">// Use ACL allocated memory</span></div> |
| 189 | <div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  }</div> |
| 190 | <div class="line"><a name="l00092"></a><span class="lineno"> 92</span>  </div> |
| 191 | <div class="line"><a name="l00093"></a><span class="lineno"> 93</span>  <span class="comment">// Create and initialize user tensors</span></div> |
| 192 | <div class="line"><a name="l00094"></a><span class="lineno"> 94</span>  std::vector<arm_compute::CLTensor*> inputsWeightsOutputs;</div> |
| 193 | <div class="line"><a name="l00095"></a><span class="lineno"> 95</span>  inputsWeightsOutputs.reserve(<a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a4b50e46a6810018f3edecfb68b2a76b3">m_Inputs</a>.size() + <a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a6abd491bb99ffe88bd472c1ae5a1ed1a">m_Outputs</a>.size());</div> |
| 194 | <div class="line"><a name="l00096"></a><span class="lineno"> 96</span>  </div> |
| 195 | <div class="line"><a name="l00097"></a><span class="lineno"> 97</span>  <span class="keywordflow">for</span> (uint32_t inputSlotIdx = 0; inputSlotIdx < <a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a4b50e46a6810018f3edecfb68b2a76b3">m_Inputs</a>.size(); ++inputSlotIdx)</div> |
| 196 | <div class="line"><a name="l00098"></a><span class="lineno"> 98</span>  {</div> |
| 197 | <div class="line"><a name="l00099"></a><span class="lineno"> 99</span>  arm_compute::CLTensor* input = <span class="keyword">new</span> arm_compute::CLTensor{};</div> |
| 198 | <div class="line"><a name="l00100"></a><span class="lineno"> 100</span>  <span class="comment">// inputTensorInfos is a ptr to a vector of ptrs, so we need to do a double dereference</span></div> |
| 199 | <div class="line"><a name="l00101"></a><span class="lineno"> 101</span>  input->allocator()->init(*((*inputTensorInfos)[inputSlotIdx]));</div> |
| 200 | <div class="line"><a name="l00102"></a><span class="lineno"> 102</span>  <span class="keyword">auto</span>* inputHandle = PolymorphicDowncast<GpuFsaTensorHandle*>(<a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a4b50e46a6810018f3edecfb68b2a76b3">m_Inputs</a>[inputSlotIdx]);</div> |
| 201 | <div class="line"><a name="l00103"></a><span class="lineno"> 103</span>  input->allocator()->import_memory(inputHandle->GetTensor().cl_buffer());</div> |
| 202 | <div class="line"><a name="l00104"></a><span class="lineno"> 104</span>  inputsWeightsOutputs.emplace_back(std::move(input));</div> |
| 203 | <div class="line"><a name="l00105"></a><span class="lineno"> 105</span>  }</div> |
| 204 | <div class="line"><a name="l00106"></a><span class="lineno"> 106</span>  <span class="comment">// Set the outputs</span></div> |
| 205 | <div class="line"><a name="l00107"></a><span class="lineno"> 107</span>  <span class="keywordflow">for</span> (uint32_t outputSlotIdx = 0; outputSlotIdx < <a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a6abd491bb99ffe88bd472c1ae5a1ed1a">m_Outputs</a>.size(); ++outputSlotIdx)</div> |
| 206 | <div class="line"><a name="l00108"></a><span class="lineno"> 108</span>  {</div> |
| 207 | <div class="line"><a name="l00109"></a><span class="lineno"> 109</span>  arm_compute::CLTensor* output = <span class="keyword">new</span> arm_compute::CLTensor{};</div> |
| 208 | <div class="line"><a name="l00110"></a><span class="lineno"> 110</span>  <span class="comment">// outputTensorInfos is a ptr to a vector of ptrs, so we need to do a double dereference</span></div> |
| 209 | <div class="line"><a name="l00111"></a><span class="lineno"> 111</span>  output->allocator()->init(*((*outputTensorInfos)[outputSlotIdx]));</div> |
| 210 | <div class="line"><a name="l00112"></a><span class="lineno"> 112</span>  <span class="keyword">auto</span>* outputHandle = PolymorphicDowncast<GpuFsaTensorHandle*>(<a class="code" href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">m_Data</a>.<a class="code" href="structarmnn_1_1_queue_descriptor.html#a6abd491bb99ffe88bd472c1ae5a1ed1a">m_Outputs</a>[outputSlotIdx]);</div> |
| 211 | <div class="line"><a name="l00113"></a><span class="lineno"> 113</span>  output->allocator()->import_memory(outputHandle->GetTensor().cl_buffer());</div> |
| 212 | <div class="line"><a name="l00114"></a><span class="lineno"> 114</span>  inputsWeightsOutputs.emplace_back(std::move(output));</div> |
| 213 | <div class="line"><a name="l00115"></a><span class="lineno"> 115</span>  }</div> |
| 214 | <div class="line"><a name="l00116"></a><span class="lineno"> 116</span>  runtime.run(inputsWeightsOutputs);</div> |
| 215 | <div class="line"><a name="l00117"></a><span class="lineno"> 117</span> }</div> |
| 216 | <div class="line"><a name="l00118"></a><span class="lineno"> 118</span> } <span class="comment">// namespace armnn</span></div> |
| 217 | </div><!-- fragment --></div><!-- contents --> |
| 218 | </div><!-- doc-content --> |
| 219 | <div class="ttc" id="astructarmnn_1_1_pre_compiled_queue_descriptor_html_aa1fedd1a6371526cb68cc5cc58c87465"><div class="ttname"><a href="structarmnn_1_1_pre_compiled_queue_descriptor.html#aa1fedd1a6371526cb68cc5cc58c87465">armnn::PreCompiledQueueDescriptor::m_PreCompiledObject</a></div><div class="ttdeci">void * m_PreCompiledObject</div><div class="ttdef"><b>Definition:</b> <a href="_workload_data_8hpp_source.html#l00519">WorkloadData.hpp:519</a></div></div> |
| 220 | <div class="ttc" id="astructarmnn_1_1_pre_compiled_queue_descriptor_html"><div class="ttname"><a href="structarmnn_1_1_pre_compiled_queue_descriptor.html">armnn::PreCompiledQueueDescriptor</a></div><div class="ttdef"><b>Definition:</b> <a href="_workload_data_8hpp_source.html#l00512">WorkloadData.hpp:512</a></div></div> |
| 221 | <div class="ttc" id="a_gpu_fsa_tensor_handle_8hpp_html"><div class="ttname"><a href="_gpu_fsa_tensor_handle_8hpp.html">GpuFsaTensorHandle.hpp</a></div></div> |
| 222 | <div class="ttc" id="astructarmnn_1_1_gpu_fsa_pre_compiled_blob_html_a479b90f0b24c427502d94b716117e034"><div class="ttname"><a href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a479b90f0b24c427502d94b716117e034">armnn::GpuFsaPreCompiledBlob::inputTensorInfos</a></div><div class="ttdeci">std::unique_ptr< std::vector< arm_compute::ITensorInfo * > > inputTensorInfos</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_backend_8hpp_source.html#l00037">GpuFsaBackend.hpp:37</a></div></div> |
| 223 | <div class="ttc" id="astructarmnn_1_1_workload_info_html"><div class="ttname"><a href="structarmnn_1_1_workload_info.html">armnn::WorkloadInfo</a></div><div class="ttdoc">Contains information about TensorInfos of a layer.</div><div class="ttdef"><b>Definition:</b> <a href="_workload_info_8hpp_source.html#l00016">WorkloadInfo.hpp:16</a></div></div> |
| 224 | <div class="ttc" id="a_polymorphic_downcast_8hpp_html"><div class="ttname"><a href="_polymorphic_downcast_8hpp.html">PolymorphicDowncast.hpp</a></div></div> |
| 225 | <div class="ttc" id="astructarmnn_1_1_gpu_fsa_pre_compiled_blob_html_a14f92a9f65e32c3da896e7b1d45abd02"><div class="ttname"><a href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a14f92a9f65e32c3da896e7b1d45abd02">armnn::GpuFsaPreCompiledBlob::sketch</a></div><div class="ttdeci">std::unique_ptr< arm_compute::experimental::dynamic_fusion::GpuWorkloadSketch > sketch</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_backend_8hpp_source.html#l00034">GpuFsaBackend.hpp:34</a></div></div> |
| 226 | <div class="ttc" id="aclassarmnn_1_1_invalid_argument_exception_html"><div class="ttname"><a href="classarmnn_1_1_invalid_argument_exception.html">armnn::InvalidArgumentException</a></div><div class="ttdef"><b>Definition:</b> <a href="_exceptions_8hpp_source.html#l00080">Exceptions.hpp:80</a></div></div> |
| 227 | <div class="ttc" id="anamespacearmnn_html_a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c"><div class="ttname"><a href="namespacearmnn.html#a4dc0adc6737b5944e7671bee71788407acaf9b6b99962bf5c2264824231d7a40c">armnn::BoostLogSeverityMapping::info</a></div><div class="ttdeci">@ info</div></div> |
| 228 | <div class="ttc" id="astructarmnn_1_1_queue_descriptor_html_a6abd491bb99ffe88bd472c1ae5a1ed1a"><div class="ttname"><a href="structarmnn_1_1_queue_descriptor.html#a6abd491bb99ffe88bd472c1ae5a1ed1a">armnn::QueueDescriptor::m_Outputs</a></div><div class="ttdeci">std::vector< ITensorHandle * > m_Outputs</div><div class="ttdef"><b>Definition:</b> <a href="_workload_data_8hpp_source.html#l00027">WorkloadData.hpp:27</a></div></div> |
| 229 | <div class="ttc" id="a_gpu_fsa_workload_utils_8hpp_html"><div class="ttname"><a href="_gpu_fsa_workload_utils_8hpp.html">GpuFsaWorkloadUtils.hpp</a></div></div> |
| 230 | <div class="ttc" id="aclassarmnn_1_1_base_workload_html"><div class="ttname"><a href="classarmnn_1_1_base_workload.html">armnn::BaseWorkload</a></div><div class="ttdef"><b>Definition:</b> <a href="_workload_8hpp_source.html#l00033">Workload.hpp:33</a></div></div> |
| 231 | <div class="ttc" id="aclassarmnn_1_1_gpu_fsa_pre_compiled_workload_html_aeef6e7488b495adc96f8abd7d29e3d5f"><div class="ttname"><a href="classarmnn_1_1_gpu_fsa_pre_compiled_workload.html#aeef6e7488b495adc96f8abd7d29e3d5f">armnn::GpuFsaPreCompiledWorkload::GpuFsaPreCompiledWorkload</a></div><div class="ttdeci">GpuFsaPreCompiledWorkload(const PreCompiledQueueDescriptor &descriptor, const WorkloadInfo &info)</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_pre_compiled_workload_8cpp_source.html#l00031">GpuFsaPreCompiledWorkload.cpp:31</a></div></div> |
| 232 | <div class="ttc" id="a_gpu_fsa_backend_8hpp_html"><div class="ttname"><a href="_gpu_fsa_backend_8hpp.html">GpuFsaBackend.hpp</a></div></div> |
| 233 | <div class="ttc" id="aclassarmnn_1_1_base_workload_html_afb8d2c8817c75de9d01a4c0e0d5c160b"><div class="ttname"><a href="classarmnn_1_1_base_workload.html#afb8d2c8817c75de9d01a4c0e0d5c160b">armnn::BaseWorkload< PreCompiledQueueDescriptor >::m_Data</a></div><div class="ttdeci">PreCompiledQueueDescriptor m_Data</div><div class="ttdef"><b>Definition:</b> <a href="_workload_8hpp_source.html#l00089">Workload.hpp:89</a></div></div> |
| 234 | <div class="ttc" id="anamespacearmnn_html"><div class="ttname"><a href="namespacearmnn.html">armnn</a></div><div class="ttdoc">Copyright (c) 2021 ARM Limited and Contributors.</div><div class="ttdef"><b>Definition:</b> <a href="01__00__quick__start_8dox_source.html#l00006">01_00_quick_start.dox:6</a></div></div> |
| 235 | <div class="ttc" id="a_arm_compute_tensor_utils_8hpp_html"><div class="ttname"><a href="_arm_compute_tensor_utils_8hpp.html">ArmComputeTensorUtils.hpp</a></div></div> |
| 236 | <div class="ttc" id="astructarmnn_1_1_gpu_fsa_pre_compiled_blob_html"><div class="ttname"><a href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html">armnn::GpuFsaPreCompiledBlob</a></div><div class="ttdoc">A structure which contains all the elements needed to execute a fused workload in the GpuFsa Backend.</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_backend_8hpp_source.html#l00032">GpuFsaBackend.hpp:32</a></div></div> |
| 237 | <div class="ttc" id="a_gpu_fsa_pre_compiled_workload_8hpp_html"><div class="ttname"><a href="_gpu_fsa_pre_compiled_workload_8hpp.html">GpuFsaPreCompiledWorkload.hpp</a></div></div> |
| 238 | <div class="ttc" id="aclassarmnn_1_1_gpu_fsa_pre_compiled_workload_html_ae071e8822437c78baea75c3aef3a263a"><div class="ttname"><a href="classarmnn_1_1_gpu_fsa_pre_compiled_workload.html#ae071e8822437c78baea75c3aef3a263a">armnn::GpuFsaPreCompiledWorkload::Execute</a></div><div class="ttdeci">void Execute() const override</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_pre_compiled_workload_8cpp_source.html#l00043">GpuFsaPreCompiledWorkload.cpp:43</a></div></div> |
| 239 | <div class="ttc" id="astructarmnn_1_1_gpu_fsa_pre_compiled_blob_html_ac49bf679a23aa84f06a6bde3440a4c40"><div class="ttname"><a href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#ac49bf679a23aa84f06a6bde3440a4c40">armnn::GpuFsaPreCompiledBlob::outputTensorInfos</a></div><div class="ttdeci">std::unique_ptr< std::vector< arm_compute::ITensorInfo * > > outputTensorInfos</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_backend_8hpp_source.html#l00038">GpuFsaBackend.hpp:38</a></div></div> |
| 240 | <div class="ttc" id="astructarmnn_1_1_queue_descriptor_html_a4b50e46a6810018f3edecfb68b2a76b3"><div class="ttname"><a href="structarmnn_1_1_queue_descriptor.html#a4b50e46a6810018f3edecfb68b2a76b3">armnn::QueueDescriptor::m_Inputs</a></div><div class="ttdeci">std::vector< ITensorHandle * > m_Inputs</div><div class="ttdef"><b>Definition:</b> <a href="_workload_data_8hpp_source.html#l00026">WorkloadData.hpp:26</a></div></div> |
| 241 | <!-- start footer part --> |
| 242 | <div id="nav-path" class="navpath"><!-- id is needed for treeview function! --> |
| 243 | <ul> |
| 244 | <li class="navelem"><a class="el" href="dir_68267d1309a1af8e8297ef4c3efbcdba.html">src</a></li><li class="navelem"><a class="el" href="dir_0f3cdec46afbc61a1ded8e1687c9c9a0.html">backends</a></li><li class="navelem"><a class="el" href="dir_dd5880bc3520e42c5318e86a9fdc97f6.html">gpuFsa</a></li><li class="navelem"><a class="el" href="dir_98e2ef7b561bcb53fb92be16bdf83115.html">workloads</a></li><li class="navelem"><a class="el" href="_gpu_fsa_pre_compiled_workload_8cpp.html">GpuFsaPreCompiledWorkload.cpp</a></li> |
| 245 | <li class="footer">Generated on Thu May 16 2024 09:31:47 for Arm NN by |
| 246 | <a href="http://www.doxygen.org/index.html"> |
| 247 | <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.17 </li> |
| 248 | </ul> |
| 249 | </div> |
| 250 | </body> |
| 251 | </html> |