1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
|
<!-- HTML header for doxygen 1.8.17-->
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "https://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 1.8.17"/>
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<title>Arm NN: src/backends/gpuFsa/layers/GpuFsaConvolution2d.cpp Source File</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>
<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" async="async" src="http://cdn.mathjax.org/mathjax/latest/MathJax.js"></script>
<link href="doxygen.css" rel="stylesheet" type="text/css" />
<link href="customdoxygen.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: 15rem; margin-top: .5rem; margin-left 13px"/>
<td id="projectalign" style="padding-left: 0.9em;">
<div id="projectname">
 <span id="projectnumber">24.05</span>
</div>
</td>
</tr>
</tbody>
</table>
</div>
<!-- end header part -->
<!-- Generated by Doxygen 1.8.17 -->
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */
var searchBox = new SearchBox("searchBox", "search",false,'Search');
/* @license-end */
</script>
<script type="text/javascript" src="menudata.js"></script>
<script type="text/javascript" src="menu.js"></script>
<script type="text/javascript">
/* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */
$(function() {
initMenu('',true,false,'search.php','Search');
$(document).ready(function() { init_search(); });
});
/* @license-end */</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">
/* @license magnet:?xt=urn:btih:cf05388f2679ee054f2beb29a391d25f4e673ac3&dn=gpl-2.0.txt GPL-v2 */
$(document).ready(function(){initNavTree('_gpu_fsa_convolution2d_8cpp_source.html',''); initResizable(); });
/* @license-end */
</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="headertitle">
<div class="title">GpuFsaConvolution2d.cpp</div> </div>
</div><!--header-->
<div class="contents">
<a href="_gpu_fsa_convolution2d_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>
<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>
<div class="line"><a name="l00003"></a><span class="lineno"> 3</span> <span class="comment">// SPDX-License-Identifier: MIT</span></div>
<div class="line"><a name="l00004"></a><span class="lineno"> 4</span> <span class="comment">//</span></div>
<div class="line"><a name="l00005"></a><span class="lineno"> 5</span>  </div>
<div class="line"><a name="l00006"></a><span class="lineno"> 6</span> <span class="preprocessor">#include "<a class="code" href="_gpu_fsa_convolution2d_8hpp.html">GpuFsaConvolution2d.hpp</a>"</span></div>
<div class="line"><a name="l00007"></a><span class="lineno"> 7</span> <span class="preprocessor">#include "<a class="code" href="_utils_gpu_fsa_8hpp.html">UtilsGpuFsa.hpp</a>"</span></div>
<div class="line"><a name="l00008"></a><span class="lineno"> 8</span>  </div>
<div class="line"><a name="l00009"></a><span class="lineno"> 9</span> <span class="preprocessor">#include <<a class="code" href="_arm_compute_tensor_utils_8hpp.html">aclCommon/ArmComputeTensorUtils.hpp</a>></span></div>
<div class="line"><a name="l00010"></a><span class="lineno"> 10</span>  </div>
<div class="line"><a name="l00011"></a><span class="lineno"> 11</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/GpuWorkloadContext.h></span></div>
<div class="line"><a name="l00012"></a><span class="lineno"> 12</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/GpuWorkloadSketch.h></span></div>
<div class="line"><a name="l00013"></a><span class="lineno"> 13</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/operators/GpuConv2d.h></span></div>
<div class="line"><a name="l00014"></a><span class="lineno"> 14</span> <span class="preprocessor">#include <arm_compute/dynamic_fusion/sketch/gpu/operators/GpuOutput.h></span></div>
<div class="line"><a name="l00015"></a><span class="lineno"> 15</span>  </div>
<div class="line"><a name="l00016"></a><span class="lineno"> 16</span> <span class="preprocessor">#include <vector></span></div>
<div class="line"><a name="l00017"></a><span class="lineno"> 17</span>  </div>
<div class="line"><a name="l00018"></a><span class="lineno"> 18</span> <span class="keyword">using namespace </span>arm_compute::experimental::dynamic_fusion;</div>
<div class="line"><a name="l00019"></a><span class="lineno"> 19</span> <span class="keyword">using namespace </span>armnn::armcomputetensorutils;</div>
<div class="line"><a name="l00020"></a><span class="lineno"> 20</span>  </div>
<div class="line"><a name="l00021"></a><span class="lineno"> 21</span> <span class="keyword">namespace </span><a class="code" href="namespacearmnn.html">armnn</a></div>
<div class="line"><a name="l00022"></a><span class="lineno"> 22</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"><a class="line" href="namespacearmnn.html#a1c42ef784299460fb9db004dd395d5e1"> 24</a></span> <a class="code" href="namespacearmnn.html#a67a0db04d321a74b7e7fcfd3f1a3f70b">arm_compute::Status</a> <a class="code" href="namespacearmnn.html#a1c42ef784299460fb9db004dd395d5e1">GpuFsaConvolution2dValidate</a>(<span class="keyword">const</span> <a class="code" href="classarmnn_1_1_tensor_info.html">TensorInfo</a>& input,</div>
<div class="line"><a name="l00025"></a><span class="lineno"> 25</span>  <span class="keyword">const</span> <a class="code" href="structarmnn_1_1_convolution2d_descriptor.html">Convolution2dDescriptor</a>& descriptor,</div>
<div class="line"><a name="l00026"></a><span class="lineno"> 26</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_tensor_info.html">TensorInfo</a>& weights,</div>
<div class="line"><a name="l00027"></a><span class="lineno"> 27</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_optional.html">Optional<TensorInfo></a>& biases)</div>
<div class="line"><a name="l00028"></a><span class="lineno"> 28</span> {</div>
<div class="line"><a name="l00029"></a><span class="lineno"> 29</span>  <span class="comment">// Create a new workload sketch, for validation purposes</span></div>
<div class="line"><a name="l00030"></a><span class="lineno"> 30</span>  <span class="keyword">auto</span> compileCtx = arm_compute::CLKernelLibrary::get().get_compile_context();</div>
<div class="line"><a name="l00031"></a><span class="lineno"> 31</span>  <span class="keyword">auto</span> workloadContext = GpuWorkloadContext(&compileCtx);</div>
<div class="line"><a name="l00032"></a><span class="lineno"> 32</span>  GpuWorkloadSketch sketch{ &workloadContext };</div>
<div class="line"><a name="l00033"></a><span class="lineno"> 33</span>  </div>
<div class="line"><a name="l00034"></a><span class="lineno"> 34</span>  <span class="comment">// Build and create tensor infos using the sketch</span></div>
<div class="line"><a name="l00035"></a><span class="lineno"> 35</span>  <span class="keyword">const</span> arm_compute::TensorInfo aclInputInfo = BuildArmComputeTensorInfo(input, descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#a6089e1ca91914015777ea780a513131a">m_DataLayout</a>);</div>
<div class="line"><a name="l00036"></a><span class="lineno"> 36</span>  arm_compute::TensorInfo aclWeightsInfo = BuildArmComputeTensorInfo(weights, descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#a6089e1ca91914015777ea780a513131a">m_DataLayout</a>);</div>
<div class="line"><a name="l00037"></a><span class="lineno"> 37</span>  aclWeightsInfo.set_are_values_constant(weights.<a class="code" href="classarmnn_1_1_tensor_info.html#a945263e85c27f3216a8323cfc16d8919">IsConstant</a>());</div>
<div class="line"><a name="l00038"></a><span class="lineno"> 38</span>  </div>
<div class="line"><a name="l00039"></a><span class="lineno"> 39</span>  <span class="keyword">auto</span> inputInfo = workloadContext.create_tensor_info(aclInputInfo);</div>
<div class="line"><a name="l00040"></a><span class="lineno"> 40</span>  <span class="keyword">auto</span> weightInfo = workloadContext.create_tensor_info(aclWeightsInfo);</div>
<div class="line"><a name="l00041"></a><span class="lineno"> 41</span>  </div>
<div class="line"><a name="l00042"></a><span class="lineno"> 42</span>  <span class="comment">// Only create the bias tensor info if enabled, otherwise pass nullptr to validate_op</span></div>
<div class="line"><a name="l00043"></a><span class="lineno"> 43</span>  arm_compute::TensorInfo aclBiasInfo;</div>
<div class="line"><a name="l00044"></a><span class="lineno"> 44</span>  arm_compute::ITensorInfo* biasSketchInfoPtr = <span class="keyword">nullptr</span>;</div>
<div class="line"><a name="l00045"></a><span class="lineno"> 45</span>  </div>
<div class="line"><a name="l00046"></a><span class="lineno"> 46</span>  <span class="keywordflow">if</span> (descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#aea202e14d8874cefd9a0f778022b7e25">m_BiasEnabled</a>)</div>
<div class="line"><a name="l00047"></a><span class="lineno"> 47</span>  {</div>
<div class="line"><a name="l00048"></a><span class="lineno"> 48</span>  <span class="keywordflow">if</span>(!biases.<a class="code" href="classarmnn_1_1_optional_base.html#a86b749ce2c4bc627fa8a1fcfaf0e314f">has_value</a>())</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="keywordflow">throw</span> <a class="code" href="classarmnn_1_1_invalid_argument_exception.html">InvalidArgumentException</a>(<span class="stringliteral">"GpuFsaConvolution2d::ValidateOp: No biases set when biases are enabled"</span>);</div>
<div class="line"><a name="l00051"></a><span class="lineno"> 51</span>  }</div>
<div class="line"><a name="l00052"></a><span class="lineno"> 52</span>  aclBiasInfo = BuildArmComputeTensorInfo(biases.<a class="code" href="classarmnn_1_1_optional_reference_switch.html#a77c7d528ac063d870b8c8426ec81c1c3">value</a>(), descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#a6089e1ca91914015777ea780a513131a">m_DataLayout</a>);</div>
<div class="line"><a name="l00053"></a><span class="lineno"> 53</span>  aclBiasInfo.set_are_values_constant(biases.<a class="code" href="classarmnn_1_1_optional_reference_switch.html#a77c7d528ac063d870b8c8426ec81c1c3">value</a>().IsConstant());</div>
<div class="line"><a name="l00054"></a><span class="lineno"> 54</span>  </div>
<div class="line"><a name="l00055"></a><span class="lineno"> 55</span>  biasSketchInfoPtr = workloadContext.create_tensor_info(aclBiasInfo);</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>  </div>
<div class="line"><a name="l00058"></a><span class="lineno"> 58</span>  Conv2dAttributes conv2dAttributes = <a class="code" href="_utils_gpu_fsa_8cpp.html#a6fb30c0d325066e78619d5e5a5611973">CreateConv2dAttributes</a>(descriptor);</div>
<div class="line"><a name="l00059"></a><span class="lineno"> 59</span>  </div>
<div class="line"><a name="l00060"></a><span class="lineno"> 60</span>  <span class="comment">// Validate operator, check status and update reasonIfUnsupported</span></div>
<div class="line"><a name="l00061"></a><span class="lineno"> 61</span>  <a class="code" href="namespacearmnn.html#a67a0db04d321a74b7e7fcfd3f1a3f70b">arm_compute::Status</a> aclStatus = GpuConv2d::validate_op(sketch,</div>
<div class="line"><a name="l00062"></a><span class="lineno"> 62</span>  inputInfo,</div>
<div class="line"><a name="l00063"></a><span class="lineno"> 63</span>  weightInfo,</div>
<div class="line"><a name="l00064"></a><span class="lineno"> 64</span>  biasSketchInfoPtr,</div>
<div class="line"><a name="l00065"></a><span class="lineno"> 65</span>  conv2dAttributes);</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>  <span class="keywordflow">return</span> aclStatus;</div>
<div class="line"><a name="l00068"></a><span class="lineno"> 68</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"><a class="line" href="namespacearmnn.html#ab8a797269fd9db3b8832998f10ad9688"> 70</a></span> <span class="keywordtype">void</span> <a class="code" href="namespacearmnn.html#ab8a797269fd9db3b8832998f10ad9688">GpuFsaConvolution2dCreateOp</a>(<a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html">GpuFsaPreCompiledBlob</a>* blob,</div>
<div class="line"><a name="l00071"></a><span class="lineno"> 71</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_tensor_info.html">TensorInfo</a>& input,</div>
<div class="line"><a name="l00072"></a><span class="lineno"> 72</span>  <span class="keyword">const</span> <a class="code" href="structarmnn_1_1_convolution2d_descriptor.html">Convolution2dDescriptor</a>& descriptor,</div>
<div class="line"><a name="l00073"></a><span class="lineno"> 73</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_tensor_info.html">TensorInfo</a>& weights,</div>
<div class="line"><a name="l00074"></a><span class="lineno"> 74</span>  <span class="keyword">const</span> <a class="code" href="classarmnn_1_1_optional.html">Optional<TensorInfo></a>& biases)</div>
<div class="line"><a name="l00075"></a><span class="lineno"> 75</span> {</div>
<div class="line"><a name="l00076"></a><span class="lineno"> 76</span> <span class="comment">/*</span></div>
<div class="line"><a name="l00077"></a><span class="lineno"> 77</span> <span class="comment"> * Creating an Op for the GpuFsa backend requires us to create and maintain quite a bit of data, which is then stored</span></div>
<div class="line"><a name="l00078"></a><span class="lineno"> 78</span> <span class="comment"> * in a GpuFsaPreCompiledBlob for execution later. Specifically we need:</span></div>
<div class="line"><a name="l00079"></a><span class="lineno"> 79</span> <span class="comment"> * GpuWorkloadContext, this contains the TensorInfos and is unique to the Graph being executed</span></div>
<div class="line"><a name="l00080"></a><span class="lineno"> 80</span> <span class="comment"> * Sketch, this is similar to a subgraph and can contain one or more operations. Multiple ops can be "fused" together</span></div>
<div class="line"><a name="l00081"></a><span class="lineno"> 81</span> <span class="comment"> * using a single sketch.</span></div>
<div class="line"><a name="l00082"></a><span class="lineno"> 82</span> <span class="comment"> * The inputTensorinfos / outputTensorInfos, these are pointers to the TensorInfos used when creating the sketch.</span></div>
<div class="line"><a name="l00083"></a><span class="lineno"> 83</span> <span class="comment"> * They refer to the TensorInfos stored within the GpuWorkloadContext and are needed when executing the sketch</span></div>
<div class="line"><a name="l00084"></a><span class="lineno"> 84</span> <span class="comment"> * as the TensorInfos used when creating the Tensors must match those used to create the Sketch. Otherwise the runtime</span></div>
<div class="line"><a name="l00085"></a><span class="lineno"> 85</span> <span class="comment"> * doesn't know which Tensors to use.</span></div>
<div class="line"><a name="l00086"></a><span class="lineno"> 86</span> <span class="comment"> */</span></div>
<div class="line"><a name="l00087"></a><span class="lineno"> 87</span>  GpuWorkloadSketch* sketch = blob-><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a14f92a9f65e32c3da896e7b1d45abd02">sketch</a>.get();</div>
<div class="line"><a name="l00088"></a><span class="lineno"> 88</span>  GpuWorkloadContext* workloadContext = blob-><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a255c9012137b149ffb46d83c23f2df43">workloadContext</a>.get();</div>
<div class="line"><a name="l00089"></a><span class="lineno"> 89</span>  std::vector<arm_compute::ITensorInfo*> inputTensorInfos = {};</div>
<div class="line"><a name="l00090"></a><span class="lineno"> 90</span>  std::vector<arm_compute::ITensorInfo*> outputTensorInfos = {};</div>
<div class="line"><a name="l00091"></a><span class="lineno"> 91</span>  </div>
<div class="line"><a name="l00092"></a><span class="lineno"> 92</span>  <span class="comment">// Build and create tensor infos using the sketch</span></div>
<div class="line"><a name="l00093"></a><span class="lineno"> 93</span>  <span class="keyword">const</span> arm_compute::TensorInfo aclInputInfo = BuildArmComputeTensorInfo(input, descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#a6089e1ca91914015777ea780a513131a">m_DataLayout</a>);</div>
<div class="line"><a name="l00094"></a><span class="lineno"> 94</span>  arm_compute::TensorInfo aclWeightsInfo = BuildArmComputeTensorInfo(weights, descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#a6089e1ca91914015777ea780a513131a">m_DataLayout</a>);</div>
<div class="line"><a name="l00095"></a><span class="lineno"> 95</span>  aclWeightsInfo.set_are_values_constant(weights.<a class="code" href="classarmnn_1_1_tensor_info.html#a945263e85c27f3216a8323cfc16d8919">IsConstant</a>());</div>
<div class="line"><a name="l00096"></a><span class="lineno"> 96</span>  </div>
<div class="line"><a name="l00097"></a><span class="lineno"> 97</span>  inputTensorInfos.emplace_back(workloadContext->create_tensor_info(aclInputInfo));</div>
<div class="line"><a name="l00098"></a><span class="lineno"> 98</span>  inputTensorInfos.emplace_back(workloadContext->create_tensor_info(aclWeightsInfo));</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>  <span class="comment">// Only create the bias tensor info if enabled, otherwise pass nullptr to validate_op / create_op</span></div>
<div class="line"><a name="l00101"></a><span class="lineno"> 101</span>  arm_compute::TensorInfo aclBiasInfo;</div>
<div class="line"><a name="l00102"></a><span class="lineno"> 102</span>  arm_compute::ITensorInfo* biasSketchInfoPtr = <span class="keyword">nullptr</span>;</div>
<div class="line"><a name="l00103"></a><span class="lineno"> 103</span>  </div>
<div class="line"><a name="l00104"></a><span class="lineno"> 104</span>  <span class="keywordflow">if</span> (descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#aea202e14d8874cefd9a0f778022b7e25">m_BiasEnabled</a>)</div>
<div class="line"><a name="l00105"></a><span class="lineno"> 105</span>  {</div>
<div class="line"><a name="l00106"></a><span class="lineno"> 106</span>  <span class="keywordflow">if</span>(!biases.<a class="code" href="classarmnn_1_1_optional_base.html#a86b749ce2c4bc627fa8a1fcfaf0e314f">has_value</a>())</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>  <span class="keywordflow">throw</span> <a class="code" href="classarmnn_1_1_invalid_argument_exception.html">InvalidArgumentException</a>(<span class="stringliteral">"GpuFsaConvolution2d::CreateOp: No biases set when biases are enabled"</span>);</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>  aclBiasInfo = BuildArmComputeTensorInfo(biases.<a class="code" href="classarmnn_1_1_optional_reference_switch.html#a77c7d528ac063d870b8c8426ec81c1c3">value</a>(), descriptor.<a class="code" href="structarmnn_1_1_convolution2d_descriptor.html#a6089e1ca91914015777ea780a513131a">m_DataLayout</a>);</div>
<div class="line"><a name="l00111"></a><span class="lineno"> 111</span>  aclBiasInfo.set_are_values_constant(biases.<a class="code" href="classarmnn_1_1_optional_reference_switch.html#a77c7d528ac063d870b8c8426ec81c1c3">value</a>().IsConstant());</div>
<div class="line"><a name="l00112"></a><span class="lineno"> 112</span>  </div>
<div class="line"><a name="l00113"></a><span class="lineno"> 113</span>  inputTensorInfos.emplace_back(workloadContext->create_tensor_info(aclBiasInfo));</div>
<div class="line"><a name="l00114"></a><span class="lineno"> 114</span>  biasSketchInfoPtr = inputTensorInfos[2];</div>
<div class="line"><a name="l00115"></a><span class="lineno"> 115</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>  Conv2dAttributes conv2dAttributes = <a class="code" href="_utils_gpu_fsa_8cpp.html#a6fb30c0d325066e78619d5e5a5611973">CreateConv2dAttributes</a>(descriptor);</div>
<div class="line"><a name="l00118"></a><span class="lineno"> 118</span>  </div>
<div class="line"><a name="l00119"></a><span class="lineno"> 119</span>  <span class="comment">// Validate operator, check status and update reasonIfUnsupported</span></div>
<div class="line"><a name="l00120"></a><span class="lineno"> 120</span>  <a class="code" href="namespacearmnn.html#a67a0db04d321a74b7e7fcfd3f1a3f70b">arm_compute::Status</a> aclStatus = GpuConv2d::validate_op(*sketch,</div>
<div class="line"><a name="l00121"></a><span class="lineno"> 121</span>  inputTensorInfos[0],</div>
<div class="line"><a name="l00122"></a><span class="lineno"> 122</span>  inputTensorInfos[1],</div>
<div class="line"><a name="l00123"></a><span class="lineno"> 123</span>  biasSketchInfoPtr,</div>
<div class="line"><a name="l00124"></a><span class="lineno"> 124</span>  conv2dAttributes);</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>  <span class="keyword">const</span> <span class="keywordtype">bool</span> supported = (aclStatus.error_code() == arm_compute::ErrorCode::OK);</div>
<div class="line"><a name="l00127"></a><span class="lineno"> 127</span>  <span class="keywordflow">if</span> (!supported)</div>
<div class="line"><a name="l00128"></a><span class="lineno"> 128</span>  {</div>
<div class="line"><a name="l00129"></a><span class="lineno"> 129</span>  <span class="keywordflow">throw</span> <a class="code" href="classarmnn_1_1_backend_capability_exception.html">BackendCapabilityException</a>(<span class="stringliteral">"\"GpuFsa\" backend failed during Convolution2D operation validation"</span>);</div>
<div class="line"><a name="l00130"></a><span class="lineno"> 130</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>  <span class="comment">// Create the Op within the Sketch using the TensorInfos we have stored</span></div>
<div class="line"><a name="l00133"></a><span class="lineno"> 133</span>  arm_compute::ITensorInfo* convOutInfo = GpuConv2d::create_op(*sketch,</div>
<div class="line"><a name="l00134"></a><span class="lineno"> 134</span>  inputTensorInfos[0],</div>
<div class="line"><a name="l00135"></a><span class="lineno"> 135</span>  inputTensorInfos[1],</div>
<div class="line"><a name="l00136"></a><span class="lineno"> 136</span>  biasSketchInfoPtr,</div>
<div class="line"><a name="l00137"></a><span class="lineno"> 137</span>  conv2dAttributes);</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>  <span class="comment">// Create the Output</span></div>
<div class="line"><a name="l00140"></a><span class="lineno"> 140</span>  outputTensorInfos.emplace_back(workloadContext->create_tensor_info());</div>
<div class="line"><a name="l00141"></a><span class="lineno"> 141</span>  GpuOutput::create_op(*sketch, convOutInfo, outputTensorInfos[0]);</div>
<div class="line"><a name="l00142"></a><span class="lineno"> 142</span>  </div>
<div class="line"><a name="l00143"></a><span class="lineno"> 143</span>  <span class="comment">// Store the TensorInfos within the blob as unique_ptrs to be used later</span></div>
<div class="line"><a name="l00144"></a><span class="lineno"> 144</span>  blob-><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a479b90f0b24c427502d94b716117e034">inputTensorInfos</a> = std::make_unique<std::vector<arm_compute::ITensorInfo*>>(inputTensorInfos);</div>
<div class="line"><a name="l00145"></a><span class="lineno"> 145</span>  blob-><a class="code" href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#ac49bf679a23aa84f06a6bde3440a4c40">outputTensorInfos</a> = std::make_unique<std::vector<arm_compute::ITensorInfo*>>(outputTensorInfos);</div>
<div class="line"><a name="l00146"></a><span class="lineno"> 146</span> }</div>
<div class="line"><a name="l00147"></a><span class="lineno"> 147</span>  </div>
<div class="line"><a name="l00148"></a><span class="lineno"> 148</span> } <span class="comment">// namespace armnn</span></div>
</div><!-- fragment --></div><!-- contents -->
</div><!-- doc-content -->
<div class="ttc" id="aclassarmnn_1_1_optional_html"><div class="ttname"><a href="classarmnn_1_1_optional.html">armnn::Optional</a></div><div class="ttdef"><b>Definition:</b> <a href="_optional_8hpp_source.html#l00270">Optional.hpp:270</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_tensor_info_html"><div class="ttname"><a href="classarmnn_1_1_tensor_info.html">armnn::TensorInfo</a></div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8hpp_source.html#l00152">Tensor.hpp:152</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_tensor_info_html_a945263e85c27f3216a8323cfc16d8919"><div class="ttname"><a href="classarmnn_1_1_tensor_info.html#a945263e85c27f3216a8323cfc16d8919">armnn::TensorInfo::IsConstant</a></div><div class="ttdeci">bool IsConstant() const</div><div class="ttdef"><b>Definition:</b> <a href="_tensor_8cpp_source.html#l00513">Tensor.cpp:513</a></div></div>
<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>
<div class="ttc" id="a_utils_gpu_fsa_8cpp_html_a6fb30c0d325066e78619d5e5a5611973"><div class="ttname"><a href="_utils_gpu_fsa_8cpp.html#a6fb30c0d325066e78619d5e5a5611973">CreateConv2dAttributes</a></div><div class="ttdeci">Conv2dAttributes CreateConv2dAttributes(const Convolution2dDescriptor &descriptor)</div><div class="ttdoc">Utility function used to setup an arm_compute::Conv2dAttributes object from given descriptor.</div><div class="ttdef"><b>Definition:</b> <a href="_utils_gpu_fsa_8cpp_source.html#l00014">UtilsGpuFsa.cpp:14</a></div></div>
<div class="ttc" id="astructarmnn_1_1_convolution2d_descriptor_html_aea202e14d8874cefd9a0f778022b7e25"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.html#aea202e14d8874cefd9a0f778022b7e25">armnn::Convolution2dDescriptor::m_BiasEnabled</a></div><div class="ttdeci">bool m_BiasEnabled</div><div class="ttdoc">Enable/disable bias.</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.html#l00582">Descriptors.hpp:582</a></div></div>
<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>
<div class="ttc" id="a_gpu_fsa_convolution2d_8hpp_html"><div class="ttname"><a href="_gpu_fsa_convolution2d_8hpp.html">GpuFsaConvolution2d.hpp</a></div></div>
<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>
<div class="ttc" id="astructarmnn_1_1_convolution2d_descriptor_html_a6089e1ca91914015777ea780a513131a"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.html#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.html#l00584">Descriptors.hpp:584</a></div></div>
<div class="ttc" id="astructarmnn_1_1_convolution2d_descriptor_html"><div class="ttname"><a href="structarmnn_1_1_convolution2d_descriptor.html">armnn::Convolution2dDescriptor</a></div><div class="ttdoc">A Convolution2dDescriptor for the Convolution2dLayer.</div><div class="ttdef"><b>Definition:</b> <a href="_descriptors_8hpp_source.html#l00534">Descriptors.hpp:534</a></div></div>
<div class="ttc" id="anamespacearmnn_html_a1c42ef784299460fb9db004dd395d5e1"><div class="ttname"><a href="namespacearmnn.html#a1c42ef784299460fb9db004dd395d5e1">armnn::GpuFsaConvolution2dValidate</a></div><div class="ttdeci">arm_compute::Status GpuFsaConvolution2dValidate(const TensorInfo &input, const Convolution2dDescriptor &descriptor, const TensorInfo &weights, const Optional< TensorInfo > &biases)</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_convolution2d_8cpp_source.html#l00024">GpuFsaConvolution2d.cpp:24</a></div></div>
<div class="ttc" id="anamespacearmnn_html_a67a0db04d321a74b7e7fcfd3f1a3f70b"><div class="ttname"><a href="namespacearmnn.html#a67a0db04d321a74b7e7fcfd3f1a3f70b">armnn::Status</a></div><div class="ttdeci">Status</div><div class="ttdef"><b>Definition:</b> <a href="_types_8hpp_source.html#l00042">Types.hpp:42</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_backend_capability_exception_html"><div class="ttname"><a href="classarmnn_1_1_backend_capability_exception.html">armnn::BackendCapabilityException</a></div><div class="ttdef"><b>Definition:</b> <a href="_exceptions_8hpp_source.html#l00152">Exceptions.hpp:152</a></div></div>
<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>
<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>
<div class="ttc" id="astructarmnn_1_1_gpu_fsa_pre_compiled_blob_html_a255c9012137b149ffb46d83c23f2df43"><div class="ttname"><a href="structarmnn_1_1_gpu_fsa_pre_compiled_blob.html#a255c9012137b149ffb46d83c23f2df43">armnn::GpuFsaPreCompiledBlob::workloadContext</a></div><div class="ttdeci">std::shared_ptr< arm_compute::experimental::dynamic_fusion::GpuWorkloadContext > workloadContext</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_backend_8hpp_source.html#l00035">GpuFsaBackend.hpp:35</a></div></div>
<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>
<div class="ttc" id="a_utils_gpu_fsa_8hpp_html"><div class="ttname"><a href="_utils_gpu_fsa_8hpp.html">UtilsGpuFsa.hpp</a></div></div>
<div class="ttc" id="anamespacearmnn_html_ab8a797269fd9db3b8832998f10ad9688"><div class="ttname"><a href="namespacearmnn.html#ab8a797269fd9db3b8832998f10ad9688">armnn::GpuFsaConvolution2dCreateOp</a></div><div class="ttdeci">void GpuFsaConvolution2dCreateOp(GpuFsaPreCompiledBlob *blob, const TensorInfo &input, const Convolution2dDescriptor &descriptor, const TensorInfo &weights, const Optional< TensorInfo > &biases)</div><div class="ttdef"><b>Definition:</b> <a href="_gpu_fsa_convolution2d_8cpp_source.html#l00070">GpuFsaConvolution2d.cpp:70</a></div></div>
<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>
<div class="ttc" id="aclassarmnn_1_1_optional_reference_switch_html_a77c7d528ac063d870b8c8426ec81c1c3"><div class="ttname"><a href="classarmnn_1_1_optional_reference_switch.html#a77c7d528ac063d870b8c8426ec81c1c3">armnn::OptionalReferenceSwitch< std::is_reference< T >::value, T >::value</a></div><div class="ttdeci">const T & value() const</div><div class="ttdef"><b>Definition:</b> <a href="_optional_8hpp_source.html#l00146">Optional.hpp:146</a></div></div>
<div class="ttc" id="aclassarmnn_1_1_optional_base_html_a86b749ce2c4bc627fa8a1fcfaf0e314f"><div class="ttname"><a href="classarmnn_1_1_optional_base.html#a86b749ce2c4bc627fa8a1fcfaf0e314f">armnn::OptionalBase::has_value</a></div><div class="ttdeci">bool has_value() const noexcept</div><div class="ttdef"><b>Definition:</b> <a href="_optional_8hpp_source.html#l00053">Optional.hpp:53</a></div></div>
<!-- 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.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_6735d107a3ea44e973a098b746f1b3bd.html">layers</a></li><li class="navelem"><a class="el" href="_gpu_fsa_convolution2d_8cpp.html">GpuFsaConvolution2d.cpp</a></li>
<li class="footer">Generated on Thu May 16 2024 09:31:47 for Arm NN by
<a href="http://www.doxygen.org/index.html">
<img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.8.17 </li>
</ul>
</div>
</body>
</html>
|
