https://calee0219.github.io/tag/driver/DriverHugoBlox Kit (https://hugoblox.com)en-us©Mon, 09 Aug 2021 07:48:37 +0800https://calee0219.github.io/media/icon_hu_da05098ef60dc2e7.pnghttps://calee0219.github.io/tag/driver/https://calee0219.github.io/blog/note_nvidia/Mon, 09 Aug 2021 07:48:37 +0800https://calee0219.github.io/blog/note_nvidia/<h2 id="nvidia-driver-on-ubuntu">Nvidia driver on Ubuntu</h2> <p> </p> <ul> <li>到 查看最新版本 <ul> <li> </li> <li>34x: 傳統架構版本</li> <li>387: 新版短期</li> <li>384: 新版長期</li> </ul> </li> <li>刪除舊版</li> </ul> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-shell" data-lang="shell"><span class="line"><span class="cl">sudo apt-get purge nvidia* </span></span></code></pre></div><ul> <li>安裝新版</li> </ul> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-shell" data-lang="shell"><span class="line"><span class="cl">sudo add-apt-repository ppa:graphics-drivers </span></span><span class="line"><span class="cl">sudo apt-get update </span></span><span class="line"><span class="cl">sudo apt-get install nvidia-387 </span></span><span class="line"><span class="cl">sudo reboot <span class="c1"># 重開機</span> </span></span></code></pre></div><ul> <li>檢查</li> </ul> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-shell" data-lang="shell"><span class="line"><span class="cl">lsmod <span class="p">|</span> grep nvidia </span></span><span class="line"><span class="cl"> </span></span><span class="line"><span class="cl">nvidia-smi </span></span></code></pre></div><h2 id="cuda">cuda</h2> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-shell" data-lang="shell"><span class="line"><span class="cl">sudo apt install cuda-9 </span></span><span class="line"><span class="cl">sudo apt install cuda-nvcc-9-1 </span></span></code></pre></div><ul> <li>手動安裝 到 下載 runfile(local)</li> </ul> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-shell" data-lang="shell"><span class="line"><span class="cl">sudo sh cuda_9.1.85_387.26_linux.run </span></span></code></pre></div><ul> <li>檢查 注意：gcc 要是 &lt; 6 的版本 @@</li> </ul> <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-shell" data-lang="shell"><span class="line"><span class="cl">cp -r /usr/local/cuda-9.1/samples ./ </span></span><span class="line"><span class="cl"><span class="nb">cd</span> samples </span></span><span class="line"><span class="cl">make </span></span></code></pre></div><h2 id="cudnn">cudnn</h2> <p>到 下載</p> <ul> <li>cuda 的資料夾要在 <code>/usr/local/cuda/</code>，so 不管裝在哪裡，都給一個 link 到 <code>/usr/local/cuda</code></li> <li>ubuntu 版: <ul> <li>下載 libcudnn7&hellip;deb, libcudnn7-dev&hellip;deb, libcudnn7-doc&hellip;deb</li> <li>三個都做 <code>sudo dpkg -i</code></li> </ul> </li> <li>其他 <ul> <li>下載壓縮檔</li> <li>解壓縮: <code>tar -xzvf cudnn-9.0-linux-x64-v7.tgz</code></li> <li>安裝: <div class="highlight"><pre tabindex="0" class="chroma"><code class="language-fallback" data-lang="fallback"><span class="line"><span class="cl">$ sudo cp cuda/include/cudnn.h /usr/local/cuda/include </span></span><span class="line"><span class="cl">$ sudo cp cuda/lib64/libcudnn* /usr/local/cuda/lib64 </span></span><span class="line"><span class="cl">$ sudo chmod a+r /usr/local/cuda/include/cudnn.h /usr/local/cuda/lib64/libcudnn* </span></span></code></pre></div></li> </ul> </li> </ul>