<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[XR806+TinyMaix，在XR806上实现ML推理]]></title><description><![CDATA[<h3>关于 TinyMaix</h3>
<p dir="auto">TinyMaix是面向单片机的超轻量级的神经网络推理库，即TinyML推理库，可以让你在任意单片机上运行轻量级深度学习模型~</p>
<p dir="auto">设计原则：易用性 &gt; 移植性 &gt; 速度 &gt; 空间</p>
<p dir="auto">核心代码少于400行(tm_layers.c+tm_model.c+arch_cpu.h), 代码段(.text)少于3KB<br />
低内存消耗，甚至Arduino ATmega328 (32KB Flash, 2KB Ram) 都能基于TinyMaix跑mnist(手写数字识别)<br />
支持INT8/FP32/FP16模型，实验性地支持FP8模型，支持keras h5或tflite模型转换<br />
支持多种芯片架构的专用指令优化: ARM SIMD/NEON/MVEI，RV32P, RV64V<br />
友好的用户接口，只需要load/run模型~<br />
支持全静态的内存配置(无需malloc)</p>
<h3>XR806 测试 mnist</h3>
<p dir="auto"><img src="/assets/uploads/files/1663653859154-6748e5ac-14ef-427f-8bc1-16242e532917-image.png" alt="6748e5ac-14ef-427f-8bc1-16242e532917-image.png" class=" img-responsive img-markdown" width="697" height="434" /></p>
<h3>TinyMaix XR806 Port</h3>
<p dir="auto">首先，仓库地址：<br />
<a href="https://github.com/YuzukiHD/TinyMaix-XR806" target="_blank" rel="noopener noreferrer nofollow ugc">https://github.com/YuzukiHD/TinyMaix-XR806</a></p>
<p dir="auto">测试跑分结果：</p>
<table class="table table-bordered table-striped">
<thead>
<tr>
<th>config</th>
<th>mnist (SRAM)</th>
<th>cifar</th>
<th>vww96 (Flash XIP) 32K Cache</th>
<th>mbnet128 (Flash XIP) 32K Cache</th>
<th>mbnet128 (Flash XIP) no Cache</th>
</tr>
</thead>
<tbody>
<tr>
<td>O0 CPU</td>
<td>2</td>
<td>134</td>
<td>459</td>
<td>844</td>
<td>42530</td>
</tr>
<tr>
<td>O1 CPU</td>
<td>1</td>
<td>104</td>
<td>453</td>
<td>712</td>
<td>35576</td>
</tr>
</tbody>
</table>
<p dir="auto"><code>make menuconfig</code> 的时候可以找到配置项</p>
<p dir="auto"><img src="/assets/uploads/files/1663653586643-f8460b78-116f-4e3c-acfa-ee6dd4bb49c7-image.png" alt="f8460b78-116f-4e3c-acfa-ee6dd4bb49c7-image.png" class=" img-responsive img-markdown" width="616" height="326" /></p>
<p dir="auto">同时也提供了一些demo可供使用<br />
<img src="/assets/uploads/files/1663653621817-11482b4c-a5bd-461c-bf8e-5cd9a2ec6412-image.png" alt="11482b4c-a5bd-461c-bf8e-5cd9a2ec6412-image.png" class=" img-responsive img-markdown" width="329" height="273" /></p>
]]></description><link>https://bbs.aw-ol.com/topic/2176/xr806-tinymaix-在xr806上实现ml推理</link><generator>RSS for Node</generator><lastBuildDate>Mon, 20 Jul 2026 09:42:39 GMT</lastBuildDate><atom:link href="https://bbs.aw-ol.com/topic/2176.rss" rel="self" type="application/rss+xml"/><pubDate>Tue, 20 Sep 2022 06:04:33 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to XR806+TinyMaix，在XR806上实现ML推理 on Tue, 20 Sep 2022 13:48:31 GMT]]></title><description><![CDATA[<p dir="auto"><a class="plugin-mentions-user plugin-mentions-a" href="https://bbs.aw-ol.com/uid/317">@yuzukitsuru</a> 666，找个时间试试。</p>
]]></description><link>https://bbs.aw-ol.com/post/10586</link><guid isPermaLink="true">https://bbs.aw-ol.com/post/10586</guid><dc:creator><![CDATA[dream]]></dc:creator><pubDate>Tue, 20 Sep 2022 13:48:31 GMT</pubDate></item></channel></rss>