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
|
<!-- 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: README.md 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>
<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">22.02</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('_r_e_a_d_m_e_8md.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="headertitle">
<div class="title">README.md</div> </div>
</div><!--header-->
<div class="contents">
<a href="_r_e_a_d_m_e_8md.xhtml">Go to the documentation of this file.</a><div class="fragment"><div class="line"><a name="l00001"></a><span class="lineno"> 1</span> # Introduction</div><div class="line"><a name="l00002"></a><span class="lineno"> 2</span> </div><div class="line"><a name="l00003"></a><span class="lineno"> 3</span> * [Quick Start Guides](#quick-start-guides)</div><div class="line"><a name="l00004"></a><span class="lineno"> 4</span> * [Software tools overview](#software-tools-overview)</div><div class="line"><a name="l00005"></a><span class="lineno"> 5</span> * [Where to find more information](#where-to-find-more-information)</div><div class="line"><a name="l00006"></a><span class="lineno"> 6</span> * [Contributions](#contributions)</div><div class="line"><a name="l00007"></a><span class="lineno"> 7</span> * [Disclaimer](#disclaimer)</div><div class="line"><a name="l00008"></a><span class="lineno"> 8</span> * [License](#license)</div><div class="line"><a name="l00009"></a><span class="lineno"> 9</span> * [Third-Party](#third-party)</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> Arm NN is a key component of the [machine learning platform](https://mlplatform.org/), which is part of the</div><div class="line"><a name="l00012"></a><span class="lineno"> 12</span> [Linaro Machine Intelligence Initiative](https://www.linaro.org/news/linaro-announces-launch-of-machine-intelligence-initiative/).</div><div class="line"><a name="l00013"></a><span class="lineno"> 13</span> </div><div class="line"><a name="l00014"></a><span class="lineno"> 14</span> The Arm NN SDK is a set of open-source software and tools that enables machine learning workloads on power-efficient</div><div class="line"><a name="l00015"></a><span class="lineno"> 15</span> devices. It provides a bridge between existing neural network frameworks and power-efficient Cortex-A CPUs,</div><div class="line"><a name="l00016"></a><span class="lineno"> 16</span> Arm Mali GPUs and Arm Ethos NPUs.</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> <img align="center" width="400" src="https://developer.arm.com/-/media/Arm Developer Community/Images/Block Diagrams/Arm-NN/Arm-NN-Frameworks-Diagram.png"/></div><div class="line"><a name="l00019"></a><span class="lineno"> 19</span> </div><div class="line"><a name="l00020"></a><span class="lineno"> 20</span> Arm NN SDK utilizes the Compute Library to target programmable cores, such as Cortex-A CPUs and Mali GPUs,</div><div class="line"><a name="l00021"></a><span class="lineno"> 21</span> as efficiently as possible. To target Ethos NPUs the NPU-Driver is utilized. We also welcome new contributors to provide</div><div class="line"><a name="l00022"></a><span class="lineno"> 22</span> their [own driver and backend](src/backends/README.md). Note, Arm NN does not provide support for Cortex-M CPUs.</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> Arm NN support models created with **TensorFlow Lite** (TfLite) and **ONNX**.</div><div class="line"><a name="l00025"></a><span class="lineno"> 25</span> Arm NN analysis a given model and replaces the operations within it with implementations particularly designed for the</div><div class="line"><a name="l00026"></a><span class="lineno"> 26</span> hardware you want to execute it on. This results in a great boost of execution speed. How much faster your neural</div><div class="line"><a name="l00027"></a><span class="lineno"> 27</span> network can be executed depends on the operations it contains and the available hardware. Below you can see the speedup</div><div class="line"><a name="l00028"></a><span class="lineno"> 28</span> we've been experiencing in our experiments with a few common networks.</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> \image html PerformanceChart.png</div><div class="line"><a name="l00031"></a><span class="lineno"> 31</span> </div><div class="line"><a name="l00032"></a><span class="lineno"> 32</span> Arm NN is written using portable C++14 and the build system uses [CMake](https://cmake.org/), therefore it is possible</div><div class="line"><a name="l00033"></a><span class="lineno"> 33</span> to build for a wide variety of target platforms, from a wide variety of host environments.</div><div class="line"><a name="l00034"></a><span class="lineno"> 34</span> </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> ## Getting started: Quick Start Guides </div><div class="line"><a name="l00037"></a><span class="lineno"> 37</span> Arm NN has added some quick start guides that will help you to setup Arm NN and run models quickly. The quickest way to build Arm NN is to either use our **Debian package** or use the prebuilt binaries available in the [Assets](https://github.com/ARM-software/armnn/releases) section of every Arm NN release. </div><div class="line"><a name="l00038"></a><span class="lineno"> 38</span> There is an installation guide available [here](InstallationViaAptRepository.md) which provides step by step instructions on how to install the Arm NN Core,</div><div class="line"><a name="l00039"></a><span class="lineno"> 39</span> the TfLite Parser and PyArmNN for Ubuntu 20.04. These guides can be used with the **prebuilt binaries**. </div><div class="line"><a name="l00040"></a><span class="lineno"> 40</span> At present we have added a [quick start guide](delegate/DelegateQuickStartGuide.md) that will show you how to integrate the delegate into TfLite to run models using python.</div><div class="line"><a name="l00041"></a><span class="lineno"> 41</span> More guides will be added here in the future.</div><div class="line"><a name="l00042"></a><span class="lineno"> 42</span> </div><div class="line"><a name="l00043"></a><span class="lineno"> 43</span> </div><div class="line"><a name="l00044"></a><span class="lineno"> 44</span> ## Software Components overview</div><div class="line"><a name="l00045"></a><span class="lineno"> 45</span> Depending on what kind of framework (Tensorflow Lite, ONNX) you've been using to create your model there are multiple</div><div class="line"><a name="l00046"></a><span class="lineno"> 46</span> software tools available within Arm NN that can serve your needs.</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> Generally, there is a **parser** available **for each supported framework**. ArmNN-Parsers are C++ libraries that you can integrate into your application to load, optimize and execute your model.</div><div class="line"><a name="l00049"></a><span class="lineno"> 49</span> Each parser allows you to run models from one framework. If you would like to run an ONNX model you can make use of the **Onnx-Parser**. There also is a parser available for TfLite models but the preferred way to execute TfLite models is using our TfLite-Delegate. We also provide **python bindings** for our parsers and the Arm NN core.</div><div class="line"><a name="l00050"></a><span class="lineno"> 50</span> We call the result **PyArmNN**. Therefore your application can be conveniently written in either C++ using the "original"</div><div class="line"><a name="l00051"></a><span class="lineno"> 51</span> Arm NN library or in Python using PyArmNN. You can find tutorials on how to setup and use our parsers in our doxygen</div><div class="line"><a name="l00052"></a><span class="lineno"> 52</span> documentation. The latest version can be found in the [wiki section](https://github.com/ARM-software/armnn/wiki/Documentation)</div><div class="line"><a name="l00053"></a><span class="lineno"> 53</span> of this repository.</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> Arm NN's software toolkit comes with the **TfLite Delegate** which can be integrated into TfLite.</div><div class="line"><a name="l00056"></a><span class="lineno"> 56</span> TfLite will then delegate operations, that can be accelerated with Arm NN, to Arm NN. Every other operation will still be</div><div class="line"><a name="l00057"></a><span class="lineno"> 57</span> executed with the usual TfLite runtime. This is our **recommended way to accelerate TfLite models**. As with our parsers</div><div class="line"><a name="l00058"></a><span class="lineno"> 58</span> there are tutorials in our doxygen documentation that can be found in the [wiki section](https://github.com/ARM-software/armnn/wiki/Documentation).</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> If you would like to use **Arm NN on Android** you can follow this guide which explains [how to build Arm NN using the AndroidNDK](BuildGuideAndroidNDK.md).</div><div class="line"><a name="l00061"></a><span class="lineno"> 61</span> But you might also want to take a look at another repository which implements a hardware abstraction layer (HAL) for</div><div class="line"><a name="l00062"></a><span class="lineno"> 62</span> Android. The repository is called [Android-NN-Driver](https://github.com/ARM-software/android-nn-driver) and when</div><div class="line"><a name="l00063"></a><span class="lineno"> 63</span> integrated into Android it will automatically run neural networks with Arm NN.</div><div class="line"><a name="l00064"></a><span class="lineno"> 64</span> </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> ## Where to find more information</div><div class="line"><a name="l00067"></a><span class="lineno"> 67</span> The section above introduces the most important components that Arm NN provides.</div><div class="line"><a name="l00068"></a><span class="lineno"> 68</span> You can find a complete list in our **doxygen documentation**. The</div><div class="line"><a name="l00069"></a><span class="lineno"> 69</span> latest version can be found in the [wiki section](https://github.com/ARM-software/armnn/wiki/Documentation) of our github</div><div class="line"><a name="l00070"></a><span class="lineno"> 70</span> repository.</div><div class="line"><a name="l00071"></a><span class="lineno"> 71</span> </div><div class="line"><a name="l00072"></a><span class="lineno"> 72</span> For FAQs and troubleshooting advice, see [FAQ.md](docs/FAQ.md)</div><div class="line"><a name="l00073"></a><span class="lineno"> 73</span> or take a look at previous [github issues](https://github.com/ARM-software/armnn/issues).</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> </div><div class="line"><a name="l00076"></a><span class="lineno"> 76</span> ## How to get involved</div><div class="line"><a name="l00077"></a><span class="lineno"> 77</span> If you would like to get involved but don't know where to start, a good place to look is in our Github Issues.</div><div class="line"><a name="l00078"></a><span class="lineno"> 78</span> </div><div class="line"><a name="l00079"></a><span class="lineno"> 79</span> Feature requests without a volunteer to implement them are closed, but have the 'Help wanted' label, these can be found</div><div class="line"><a name="l00080"></a><span class="lineno"> 80</span> [here](https://github.com/ARM-software/armnn/issues?q=is%3Aissue+label%3A%22Help+wanted%22+).</div><div class="line"><a name="l00081"></a><span class="lineno"> 81</span> Once you find a suitable Issue, feel free to re-open it and add a comment,</div><div class="line"><a name="l00082"></a><span class="lineno"> 82</span> so that other people know you are working on it and can help.</div><div class="line"><a name="l00083"></a><span class="lineno"> 83</span> </div><div class="line"><a name="l00084"></a><span class="lineno"> 84</span> When the feature is implemented the 'Help wanted' label will be removed.</div><div class="line"><a name="l00085"></a><span class="lineno"> 85</span> </div><div class="line"><a name="l00086"></a><span class="lineno"> 86</span> ## Contributions</div><div class="line"><a name="l00087"></a><span class="lineno"> 87</span> The Arm NN project welcomes contributions. For more details on contributing to Arm NN see the [Contributing page](https://mlplatform.org/contributing/)</div><div class="line"><a name="l00088"></a><span class="lineno"> 88</span> on the [MLPlatform.org](https://mlplatform.org/) website, or see the [Contributor Guide](ContributorGuide.md).</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> Particularly if you'd like to implement your own backend next to our CPU, GPU and NPU backends there are guides for</div><div class="line"><a name="l00091"></a><span class="lineno"> 91</span> backend development:</div><div class="line"><a name="l00092"></a><span class="lineno"> 92</span> [Backend development guide](src/backends/README.md), [Dynamic backend development guide](src/dynamic/README.md)</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> </div><div class="line"><a name="l00095"></a><span class="lineno"> 95</span> ## Disclaimer</div><div class="line"><a name="l00096"></a><span class="lineno"> 96</span> The armnn/tests directory contains tests used during Arm NN development. Many of them depend on third-party IP, model</div><div class="line"><a name="l00097"></a><span class="lineno"> 97</span> protobufs and image files not distributed with Arm NN. The dependencies of some of the tests are available freely on</div><div class="line"><a name="l00098"></a><span class="lineno"> 98</span> the Internet, for those who wish to experiment, but they won't run out of the box.</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> </div><div class="line"><a name="l00101"></a><span class="lineno"> 101</span> ## License</div><div class="line"><a name="l00102"></a><span class="lineno"> 102</span> Arm NN is provided under the [MIT](https://spdx.org/licenses/MIT.html) license.</div><div class="line"><a name="l00103"></a><span class="lineno"> 103</span> See [LICENSE](LICENSE) for more information. Contributions to this project are accepted under the same license.</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> Individual files contain the following tag instead of the full license text.</div><div class="line"><a name="l00106"></a><span class="lineno"> 106</span> </div><div class="line"><a name="l00107"></a><span class="lineno"> 107</span>  SPDX-License-Identifier: MIT</div><div class="line"><a name="l00108"></a><span class="lineno"> 108</span> </div><div class="line"><a name="l00109"></a><span class="lineno"> 109</span> This enables machine processing of license information based on the SPDX License Identifiers that are available here: http://spdx.org/licenses/</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> </div><div class="line"><a name="l00112"></a><span class="lineno"> 112</span> ## Third-party</div><div class="line"><a name="l00113"></a><span class="lineno"> 113</span> Third party tools used by Arm NN:</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> | Tool | License (SPDX ID) | Description | Version | Provenience</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> | cxxopts | MIT | A lightweight C++ option parser library | SHA 12e496da3d486b87fa9df43edea65232ed852510 | https://github.com/jarro2783/cxxopts</div><div class="line"><a name="l00118"></a><span class="lineno"> 118</span> | doctest | MIT | Header-only C++ testing framework | 2.4.6 | https://github.com/onqtam/doctest</div><div class="line"><a name="l00119"></a><span class="lineno"> 119</span> | fmt | MIT | {fmt} is an open-source formatting library providing a fast and safe alternative to C stdio and C++ iostreams. | 7.0.1 | https://github.com/fmtlib/fmt</div><div class="line"><a name="l00120"></a><span class="lineno"> 120</span> | ghc | MIT | A header-only single-file std::filesystem compatible helper library | 1.3.2 | https://github.com/gulrak/filesystem</div><div class="line"><a name="l00121"></a><span class="lineno"> 121</span> | half | MIT | IEEE 754 conformant 16-bit half-precision floating point library | 1.12.0 | http://half.sourceforge.net</div><div class="line"><a name="l00122"></a><span class="lineno"> 122</span> | mapbox/variant | BSD | A header-only alternative to 'boost::variant' | 1.1.3 | https://github.com/mapbox/variant</div><div class="line"><a name="l00123"></a><span class="lineno"> 123</span> | stb | MIT | Image loader, resize and writer | 2.16 | https://github.com/nothings/stb</div><div class="line"><a name="l00124"></a><span class="lineno"> 124</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> ## Build process</div><div class="line"><a name="l00127"></a><span class="lineno"> 127</span> Arm NN uses the following security related build flags in their code:</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> | Build flags |</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> | -Wall |</div><div class="line"><a name="l00132"></a><span class="lineno"> 132</span> | -Wextra |</div><div class="line"><a name="l00133"></a><span class="lineno"> 133</span> | -Wold-style-cast |</div><div class="line"><a name="l00134"></a><span class="lineno"> 134</span> | -Wno-missing-braces |</div><div class="line"><a name="l00135"></a><span class="lineno"> 135</span> | -Wconversion |</div><div class="line"><a name="l00136"></a><span class="lineno"> 136</span> | -Wsign-conversion |</div><div class="line"><a name="l00137"></a><span class="lineno"> 137</span> | -Werror |</div></div><!-- fragment --></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="_r_e_a_d_m_e_8md.xhtml">README.md</a></li>
<li class="footer">Generated on Wed Mar 9 2022 12:01:00 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>
|
