{"id":105774,"date":"2026-09-15T23:40:40","date_gmt":"2026-09-15T15:40:40","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/105774.html"},"modified":"2026-09-15T23:40:40","modified_gmt":"2026-09-15T15:40:40","slug":"%e5%a6%82%e4%bd%95%e5%9c%a8-ubuntu-22-04-%e4%b8%8a%e5%88%a9%e7%94%a8-cuda-%e4%b8%8e-tensorrt-%e5%8a%a0%e9%80%9f-ai-%e6%8e%a8%e7%90%86%ef%bc%8c%e4%bc%98%e5%8c%96%e5%a4%9a%e6%98%be%e5%8d%a1%e6%9c%8d","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/105774.html","title":{"rendered":"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f"},"content":{"rendered":"<p>\u6211\u6709\u4e00\u53f0\u57fa\u4e8e Ubuntu 22.04 \u7684\u591a\u663e\u5361\u670d\u52a1\u5668&#xff0c;\u7528\u4e8e\u670d\u52a1\u5b9e\u65f6\u63a8\u7406\u8bf7\u6c42\u3002\u4e1a\u52a1\u573a\u666f\u5305\u62ec\u5927\u89c4\u6a21\u56fe\u50cf\u5206\u7c7b\u3001\u5bf9\u8c61\u68c0\u6d4b\u4e0e LLM&#xff08;\u5927\u8bed\u8a00\u6a21\u578b&#xff09;\u63a8\u7406&#xff0c;\u8fd9\u4e9b\u6a21\u578b\u5bf9\u5ef6\u8fdf\u548c\u541e\u5410\u6709\u6781\u9ad8\u8981\u6c42\u3002<\/p>\n<p>\u521d\u671f&#xff0c;A5\u6570\u636e\u9047\u5230\u8fc7\u8bf8\u5982&#xff1a;\u63a8\u7406\u65f6\u5ef6\u8fdf\u6ce2\u52a8\u3001\u663e\u5b58\u5229\u7528\u4e0d\u5747\u3001\u9a71\u52a8\u4e0e\u5e93\u7248\u672c\u51b2\u7a81\u3001\u4ee5\u53ca\u591a GPU \u62a2\u5360\u5bfc\u81f4\u7684\u670d\u52a1\u7a33\u5b9a\u6027\u95ee\u9898\u3002\u5728\u65e0 TensorRT \u4f18\u5316\u7684\u60c5\u51b5\u4e0b&#xff0c;\u5b9e\u9645\u541e\u5410\u6bd4\u9884\u671f\u4f4e 25% \u4ee5\u4e0a&#xff0c;\u5e76\u4f34\u968f\u663e\u5b58\u788e\u7247\u5316\u73b0\u8c61\u3002\u6700\u7ec8\u6211\u4eec\u901a\u8fc7\u5408\u7406\u6784\u5efa CUDA &#043; TensorRT \u73af\u5883\u3001\u7cbe\u7ec6\u5316\u8c03\u53c2\u548c\u8d44\u6e90\u8c03\u5ea6&#xff0c;\u5b9e\u73b0\u4e86\u663e\u8457\u7684\u6027\u80fd\u63d0\u5347\u548c\u7a33\u5b9a\u6027\u4fdd\u8bc1\u3002<\/p>\n<p>\u4ee5\u4e0b\u5185\u5bb9\u662f\u6211\u5728\u8be5\u9879\u76ee\u4e2d\u7684\u5b9e\u64cd\u603b\u7ed3&#xff0c;\u5305\u542b\u8be6\u7ec6\u6b65\u9aa4\u3001\u53c2\u6570\u3001\u4ee3\u7801\u53ca\u8bc4\u6d4b\u8868&#xff0c;\u9002\u5408\u5728 Ubuntu 22.04 \u4e0a\u642d\u5efa\u5de5\u4e1a\u7ea7 AI \u63a8\u7406\u90e8\u7f72\u73af\u5883\u3002<\/p>\n<hr \/>\n<h3>\u4e00\u3001\u786c\u4ef6\u4e0e\u7cfb\u7edf\u57fa\u7840\u914d\u7f6e<\/h3>\n<p>\u672c\u5b9e\u4f8b\u91c7\u7528\u4ee5\u4e0b\u663e\u5361\u670d\u52a1\u5668www.a5idc.com\u786c\u4ef6&#xff1a;<\/p>\n<table>\n<tr>\u90e8\u4ef6\u578b\u53f7\/\u53c2\u6570<\/tr>\n<tbody>\n<tr>\n<td>CPU<\/td>\n<td>Intel Xeon Platinum 8360Y<\/td>\n<\/tr>\n<tr>\n<td>\u4e3b\u677f<\/td>\n<td>Supermicro X12SPA-TF<\/td>\n<\/tr>\n<tr>\n<td>\u5185\u5b58<\/td>\n<td>256 GB DDR4 ECC<\/td>\n<\/tr>\n<tr>\n<td>GPU<\/td>\n<td>4 \u00d7 NVIDIA A40 48GB<\/td>\n<\/tr>\n<tr>\n<td>\u5b58\u50a8<\/td>\n<td>2 \u00d7 2TB NVMe SSD (RAID1)<\/td>\n<\/tr>\n<tr>\n<td>\u7f51\u7edc<\/td>\n<td>10 Gbps Ethernet<\/td>\n<\/tr>\n<tr>\n<td>\u64cd\u4f5c\u7cfb\u7edf<\/td>\n<td>Ubuntu 22.04.3 LTS (kernel 5.15.x)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>GPU A40 \u5c5e\u4e8e\u6570\u636e\u4e2d\u5fc3\u7ea7\u663e\u5361&#xff0c;\u9002\u5408\u9ad8\u5e76\u53d1\u63a8\u7406\u573a\u666f\u3002<\/p>\n<hr \/>\n<h3>\u4e8c\u3001\u9a71\u52a8\u3001CUDA \u4e0e TensorRT \u73af\u5883\u642d\u5efa<\/h3>\n<h4>2.1 \u5b89\u88c5 NVIDIA \u9a71\u52a8\u4e0e CUDA<\/h4>\n<li>\n<p>\u786e\u8ba4\u786c\u4ef6\u8bc6\u522b\u4e0e\u63a8\u8350\u9a71\u52a8<\/p>\n<p> <span class=\"token function\">sudo<\/span> ubuntu-drivers devices\n <\/li>\n<li>\n<p>\u5b89\u88c5 NVIDIA \u9a71\u52a8<\/p>\n<p>\u5bf9\u5e94 CUDA 12.x \u901a\u5e38\u9700\u8981 nvidia-driver-525 \u6216\u4ee5\u4e0a&#xff1a;<\/p>\n<p> <span class=\"token function\">sudo<\/span> <span class=\"token function\">apt<\/span> <span class=\"token function\">install<\/span> nvidia-driver-535<br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">reboot<\/span>\n <\/li>\n<li>\n<p>\u68c0\u67e5\u9a71\u52a8\u4e0e GPU<\/p>\n<p> nvidia-smi<\/p>\n<p>\u8f93\u51fa\u7c7b\u4f3c&#xff1a;<\/p>\n<p> &#043;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8211;&#043;<br \/>\n| NVIDIA-SMI 535.54       Driver Version: 535.54       CUDA Version: 12.9     |<br \/>\n&#043;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8212;&#8211;&#043;\n <\/li>\n<li>\n<p>\u5b89\u88c5 CUDA Toolkit<\/p>\n<p>\u5efa\u8bae\u4f7f\u7528 NVIDIA \u5b98\u65b9 repo&#xff1a;<\/p>\n<p> <span class=\"token function\">sudo<\/span> <span class=\"token function\">apt<\/span> update<br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">apt<\/span> <span class=\"token function\">install<\/span> cuda-toolkit-12-8<\/p>\n<p>\u5b89\u88c5\u5b8c\u6210\u540e\u914d\u7f6e\u73af\u5883\u53d8\u91cf&#xff1a;<\/p>\n<p> <span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#039;export PATH&#061;\/usr\/local\/cuda-12.8\/bin:$PATH&#039;<\/span> <span class=\"token operator\">&gt;&gt;<\/span> ~\/.bashrc<br \/>\n<span class=\"token builtin class-name\">echo<\/span> <span class=\"token string\">&#039;export LD_LIBRARY_PATH&#061;\/usr\/local\/cuda-12.8\/lib64:$LD_LIBRARY_PATH&#039;<\/span> <span class=\"token operator\">&gt;&gt;<\/span> ~\/.bashrc<br \/>\n<span class=\"token builtin class-name\">source<\/span> ~\/.bashrc\n <\/li>\n<p>\u662f\u5426\u9009\u7528 CUDA 12.8 \/ 12.9&#xff0c;\u5e94\u6839\u636e TensorRT \u517c\u5bb9\u6027\u6765\u5b9a&#xff08;TensorRT \u5bf9 CUDA 12.x \u7cfb\u5217\u6709\u652f\u6301&#xff09;&#xff0c;\u907f\u514d\u7248\u672c\u4e0d\u5339\u914d\u5bfc\u81f4\u5b89\u88c5\u5931\u8d25\u6216\u8fd0\u884c\u9519\u8bef\u3002<\/p>\n<hr \/>\n<h4>2.2 \u5b89\u88c5 cuDNN \u4e0e TensorRT<\/h4>\n<p>TensorRT \u662f NVIDIA \u4e13\u4e3a\u63a8\u7406\u4f18\u5316\u7684 SDK&#xff0c;\u80fd\u5c06\u5e38\u89c4\u8bad\u7ec3\u6a21\u578b\u8f6c\u6362\u6210\u9ad8\u6548\u7684\u4f18\u5316\u5f15\u64ce\u3002\u53ef\u4ece NVIDIA Developer \u4e0b\u8f7d\u5bf9\u5e94 CUDA \u7684 TensorRT \u5b89\u88c5\u5305\u3002<\/p>\n<li>\n<p>\u4e0b\u8f7d TensorRT<\/p>\n<p>\u5728\u5f00\u53d1\u8005\u5b98\u7f51\u9009\u62e9\u5bf9\u5e94 CUDA \u7248\u672c\u7684 TensorRT \u5305&#xff0c;\u5982 TensorRT 10.x for CUDA 12.x\u3002<\/p>\n<\/li>\n<li>\n<p>\u89e3\u538b\u5e76\u5b89\u88c5<\/p>\n<p> <span class=\"token function\">tar<\/span> -xf TensorRT-10.14.1.48.Linux.x86_64-gnu.cuda-12.9.tar.gz<br \/>\n<span class=\"token function\">sudo<\/span> <span class=\"token function\">cp<\/span> -r TensorRT-10.14.1.48\/* \/usr\/local\/TensorRT-10.14.1.48\/\n <\/li>\n<li>\n<p>\u914d\u7f6e\u73af\u5883\u53d8\u91cf<\/p>\n<p> <span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">TRT_LIB<\/span><span class=\"token operator\">&#061;<\/span>\/usr\/local\/TensorRT-10.14.1.48\/lib<br \/>\n<span class=\"token builtin class-name\">export<\/span> <span class=\"token assign-left variable\">LD_LIBRARY_PATH<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token variable\">$TRT_LIB<\/span><span class=\"token builtin class-name\">:<\/span><span class=\"token variable\">$LD_LIBRARY_PATH<\/span>\n <\/li>\n<li>\n<p>\u786e\u8ba4 TensorRT \u5b89\u88c5<\/p>\n<p>\u7b80\u5355\u8fd0\u884c&#xff1a;<\/p>\n<p> dpkg -l <span class=\"token operator\">|<\/span> <span class=\"token function\">grep<\/span> nvinfer<\/p>\n<p>\u5982\u679c\u663e\u793a TensorRT \u6838\u5fc3\u5305&#xff0c;\u5219\u5b89\u88c5\u6210\u529f\u3002<\/p>\n<\/li>\n<hr \/>\n<h3>\u4e09\u3001\u6784\u5efa\u63a8\u7406\u670d\u52a1\u73af\u5883\u4e0e\u4f9d\u8d56<\/h3>\n<p>\u4e3a\u4e86\u9694\u79bb\u73af\u5883\u5e76\u4fbf\u4e8e\u7ba1\u7406\u4f9d\u8d56&#xff0c;\u5efa\u8bae\u4f7f\u7528 Conda \u6216 virtualenv&#xff1a;<\/p>\n<p>conda create -n ai-infer <span class=\"token assign-left variable\">python<\/span><span class=\"token operator\">&#061;<\/span><span class=\"token number\">3.10<\/span><br \/>\nconda activate ai-infer<br \/>\npip <span class=\"token function\">install<\/span> numpy onnxruntime-gpu<\/p>\n<p>\u5bf9\u4e8e ONNX &#043; TensorRT \u63a8\u7406&#xff0c;\u6211\u4eec\u9700\u8981\u5b89\u88c5 onnx \u53ca\u5bf9\u5e94\u7684 ONNX parser&#xff1a;<\/p>\n<p>pip <span class=\"token function\">install<\/span> onnx onnx-tensorrt<\/p>\n<hr \/>\n<h3>\u56db\u3001TensorRT \u63a8\u7406\u52a0\u901f\u5b9e\u73b0<\/h3>\n<h4>4.1 ONNX \u6a21\u578b\u8f6c\u6362\u4e3a TensorRT \u5f15\u64ce<\/h4>\n<p>\u4ee5\u4e0b\u793a\u4f8b\u662f\u57fa\u4e8e\u4e00\u4e2a ResNet50 ONNX \u6a21\u578b\u7684 TensorRT \u5f15\u64ce\u6784\u5efa\u6d41\u7a0b&#xff1a;<\/p>\n<p><span class=\"token keyword\">import<\/span> tensorrt <span class=\"token keyword\">as<\/span> trt<\/p>\n<p>TRT_LOGGER <span class=\"token operator\">&#061;<\/span> trt<span class=\"token punctuation\">.<\/span>Logger<span class=\"token punctuation\">(<\/span>trt<span class=\"token punctuation\">.<\/span>Logger<span class=\"token punctuation\">.<\/span>WARNING<span class=\"token punctuation\">)<\/span><br \/>\nbuilder <span class=\"token operator\">&#061;<\/span> trt<span class=\"token punctuation\">.<\/span>Builder<span class=\"token punctuation\">(<\/span>TRT_LOGGER<span class=\"token punctuation\">)<\/span><br \/>\nnetwork <span class=\"token operator\">&#061;<\/span> builder<span class=\"token punctuation\">.<\/span>create_network<span class=\"token punctuation\">(<\/span><span class=\"token number\">1<\/span> <span class=\"token operator\">&lt;&lt;<\/span> <span class=\"token builtin\">int<\/span><span class=\"token punctuation\">(<\/span>trt<span class=\"token punctuation\">.<\/span>NetworkDefinitionCreationFlag<span class=\"token punctuation\">.<\/span>EXPLICIT_BATCH<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>parser <span class=\"token operator\">&#061;<\/span> trt<span class=\"token punctuation\">.<\/span>OnnxParser<span class=\"token punctuation\">(<\/span>network<span class=\"token punctuation\">,<\/span> TRT_LOGGER<span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">with<\/span> <span class=\"token builtin\">open<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;resnet50.onnx&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;rb&#034;<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">as<\/span> model<span class=\"token punctuation\">:<\/span><br \/>\n    parser<span class=\"token punctuation\">.<\/span>parse<span class=\"token punctuation\">(<\/span>model<span class=\"token punctuation\">.<\/span>read<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>builder<span class=\"token punctuation\">.<\/span>max_workspace_size <span class=\"token operator\">&#061;<\/span> <span class=\"token number\">1<\/span> <span class=\"token operator\">&lt;&lt;<\/span> <span class=\"token number\">30<\/span><br \/>\nbuilder<span class=\"token punctuation\">.<\/span>fp16_mode <span class=\"token operator\">&#061;<\/span> <span class=\"token boolean\">True<\/span><\/p>\n<p>engine <span class=\"token operator\">&#061;<\/span> builder<span class=\"token punctuation\">.<\/span>build_cuda_engine<span class=\"token punctuation\">(<\/span>network<span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">with<\/span> <span class=\"token builtin\">open<\/span><span class=\"token punctuation\">(<\/span><span class=\"token string\">&#034;resnet50.engine&#034;<\/span><span class=\"token punctuation\">,<\/span> <span class=\"token string\">&#034;wb&#034;<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">as<\/span> f<span class=\"token punctuation\">:<\/span><br \/>\n    f<span class=\"token punctuation\">.<\/span>write<span class=\"token punctuation\">(<\/span>engine<span class=\"token punctuation\">.<\/span>serialize<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>\u6b64\u5904\u542f\u7528\u4e86 FP16 \u6a21\u5f0f\u4ee5\u63d0\u5347\u63a8\u7406\u6027\u80fd&#xff08;\u5047\u8bbe\u786c\u4ef6\u652f\u6301 FP16&#xff09;\u3002<\/p>\n<hr \/>\n<h4>4.2 \u591a\u663e\u5361 Infer Server \u5b9e\u73b0<\/h4>\n<p>\u4f7f\u7528 CUDA \u8fdb\u884c\u591a GPU \u4efb\u52a1\u5206\u914d&#xff1a;<\/p>\n<p><span class=\"token keyword\">import<\/span> pycuda<span class=\"token punctuation\">.<\/span>driver <span class=\"token keyword\">as<\/span> cuda<\/p>\n<p>cuda<span class=\"token punctuation\">.<\/span>init<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\nnum_gpus <span class=\"token operator\">&#061;<\/span> cuda<span class=\"token punctuation\">.<\/span>Device<span class=\"token punctuation\">.<\/span>count<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n<span class=\"token keyword\">for<\/span> i <span class=\"token keyword\">in<\/span> <span class=\"token builtin\">range<\/span><span class=\"token punctuation\">(<\/span>num_gpus<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span><br \/>\n    ctx <span class=\"token operator\">&#061;<\/span> cuda<span class=\"token punctuation\">.<\/span>Device<span class=\"token punctuation\">(<\/span>i<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>make_context<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><br \/>\n    <span class=\"token comment\"># \u52a0\u8f7d\u5bf9\u5e94 GPU \u4e0a\u7684 TensorRT \u5f15\u64ce\u7b49<\/span><br \/>\n    ctx<span class=\"token punctuation\">.<\/span>pop<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><\/p>\n<p>\u5408\u7406\u5206\u914d\u5f15\u64ce\u5230\u5404 GPU \u80fd\u907f\u514d\u8de8 GPU \u540c\u6b65\u74f6\u9888\u3002<\/p>\n<hr \/>\n<h3>\u4e94\u3001\u6027\u80fd\u8c03\u4f18\u4e0e\u7a33\u5b9a\u6027\u65b9\u6848<\/h3>\n<h4>5.1 \u663e\u5b58\u9884\u70ed\u4e0e\u5185\u5b58\u6c60\u4f18\u5316<\/h4>\n<p>\u5728\u9ad8\u5e76\u53d1\u573a\u666f\u4e0b\u907f\u514d\u663e\u5b58\u9891\u7e41\u5206\u914d&#xff1a;<\/p>\n<p>builder<span class=\"token punctuation\">.<\/span><span class=\"token function\">set_memory_pool_limit<\/span><span class=\"token punctuation\">(<\/span>trt<span class=\"token punctuation\">.<\/span>MemoryPoolType<span class=\"token punctuation\">.<\/span>WORKSPACE<span class=\"token punctuation\">,<\/span> <span class=\"token number\">1<\/span> <span class=\"token operator\">&lt;&lt;<\/span> <span class=\"token number\">30<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">;<\/span><\/p>\n<hr \/>\n<h3>\u516d\u3001\u8bc4\u6d4b\u5bf9\u6bd4\u6570\u636e<\/h3>\n<p>\u6211\u4eec\u5bf9\u6bd4\u4e86\u672a\u4f7f\u7528 TensorRT \u4e0e\u4f7f\u7528 TensorRT \u63a8\u7406\u7684\u7ed3\u679c&#xff08;\u5355\u4f4d&#xff1a;ms&#xff09;&#xff1a;<\/p>\n<table>\n<tr>\u914d\u7f6e\u5355\u5361 1 \u5e76\u53d1\u5355\u5361 8 \u5e76\u53d14 \u5361 32 \u5e76\u53d1<\/tr>\n<tbody>\n<tr>\n<td>\u672a\u542f\u7528 TensorRT<\/td>\n<td>25.8<\/td>\n<td>78.4<\/td>\n<td>310.2<\/td>\n<\/tr>\n<tr>\n<td>\u542f\u7528 TensorRT &#043; FP16<\/td>\n<td>12.5<\/td>\n<td>38.7<\/td>\n<td>150.6<\/td>\n<\/tr>\n<tr>\n<td>TensorRT &#043; Multi GPU \u8c03\u5ea6<\/td>\n<td>12.8<\/td>\n<td>35.2<\/td>\n<td>142.3<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>\u4ece\u8868\u4e2d\u53ef\u89c1&#xff0c;\u542f\u7528 TensorRT \u540e\u5ef6\u8fdf\u964d\u4f4e\u7ea6 50%&#xff0c;\u5e76\u4e14\u591a\u663e\u5361\u8c03\u5ea6\u4e0e\u5185\u5b58\u6c60\u4f18\u5316\u80fd\u8fdb\u4e00\u6b65\u63d0\u5347\u7a33\u5b9a\u6027\u4e0e\u9ad8\u5e76\u53d1\u8868\u73b0\u3002<\/p>\n<hr \/>\n<h3>\u4e03\u3001\u8fd0\u7ef4\u7ea7\u76d1\u63a7\u4e0e\u7a33\u5b9a\u6027\u4fdd\u969c<\/h3>\n<h4>7.1 GPU \u5229\u7528\u7387\u4e0e\u9519\u8bef\u65e5\u5fd7\u76d1\u63a7<\/h4>\n<p>\u5229\u7528 nvidia-smi \u4e0e Prometheus &#043; Grafana \u76d1\u63a7 GPU \u5229\u7528\u7387\u3001\u663e\u5b58\u4f7f\u7528\u3001\u6e29\u5ea6\u4e0e\u8fdb\u7a0b\u72b6\u6001\u3002<\/p>\n<h4>7.2 \u670d\u52a1\u5bb9\u9519\u4e0e\u91cd\u542f\u7b56\u7565<\/h4>\n<p>\u4f7f\u7528 systemd \u914d\u7f6e\u63a8\u7406\u670d\u52a1\u81ea\u52a8\u91cd\u542f\u4e0e\u9519\u8bef\u9694\u79bb&#xff0c;\u4ee5\u63d0\u9ad8\u5e73\u53f0\u7a33\u5b9a\u6027&#xff1a;<\/p>\n<p>[Service]<br \/>\nExecStart&#061;\/usr\/bin\/python3 \/opt\/infer_service.py<br \/>\nRestart&#061;on-failure<br \/>\nRestartSec&#061;5<\/p>\n<hr \/>\n<h3>\u7ed3\u8bed<\/h3>\n<p>A5\u6570\u636e\u901a\u8fc7\u7cfb\u7edf\u5316\u7684 CUDA &#043; TensorRT \u90e8\u7f72\u4e0e\u8c03\u4f18&#xff0c;\u6211\u4eec\u5728 Ubuntu 22.04 \u591a GPU \u63a8\u7406\u670d\u52a1\u5668\u4e0a\u5b9e\u73b0\u4e86\u663e\u8457\u7684\u6027\u80fd\u63d0\u5347\u4e0e\u7a33\u5b9a\u6027\u4fdd\u8bc1\u3002\u672c\u65b9\u6848\u7684\u6838\u5fc3\u8981\u70b9\u5305\u62ec&#xff1a;\u5408\u7406\u7248\u672c\u9009\u62e9\u3001\u663e\u5b58\u4e0e\u5e76\u53d1\u8c03\u5ea6\u3001\u591a GPU \u8d44\u6e90\u9694\u79bb\u3001\u4ee5\u53ca\u5168\u9762\u7684\u76d1\u63a7\u548c\u5bb9\u9519\u673a\u5236\u3002\u8fd9\u5957\u65b9\u6848\u4e0d\u4ec5\u9002\u7528\u4e8e ResNet\u3001YOLO \u7b49 CV \u6a21\u578b&#xff0c;\u4e5f\u9002\u7528\u4e8e\u5927\u578b LLM \u7684\u63a8\u7406\u52a0\u901f\u3002\u5e0c\u671b\u672c\u6559\u7a0b\u80fd\u4e3a\u5b9e\u9645\u90e8\u7f72\u63d0\u4f9b\u5207\u5b9e\u53ef\u884c\u7684\u53c2\u8003\u4e0e\u542f\u53d1\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u6211\u6709\u4e00\u53f0\u57fa\u4e8e Ubuntu 22.04 \u7684\u591a\u663e\u5361\u670d\u52a1\u5668&#xff0c;\u7528\u4e8e\u670d\u52a1\u5b9e\u65f6\u63a8\u7406\u8bf7\u6c42\u3002\u4e1a\u52a1\u573a\u666f\u5305\u62ec\u5927\u89c4\u6a21\u56fe\u50cf\u5206\u7c7b\u3001\u5bf9\u8c61\u68c0\u6d4b\u4e0e LLM&#xff08;\u5927\u8bed\u8a00\u6a21\u578b&#xff09;\u63a8\u7406&#xff0c;\u8fd9\u4e9b\u6a21\u578b\u5bf9\u5ef6\u8fdf\u548c\u541e\u5410\u6709\u6781\u9ad8\u8981\u6c42\u3002<br \/>\n\u521d\u671f&#xff0c;A5\u6570\u636e\u9047\u5230\u8fc7\u8bf8\u5982&#xff1a;\u63a8\u7406\u65f6\u5ef6\u8fdf\u6ce2\u52a8\u3001\u663e\u5b58\u5229\u7528\u4e0d\u5747\u3001\u9a71\u52a8\u4e0e\u5e93\u7248\u672c\u51b2\u7a81\u3001\u4ee5\u53ca\u591a GPU \u62a2\u5360\u5bfc\u81f4\u7684\u670d\u52a1\u7a33\u5b9a\u6027\u95ee\u9898\u3002\u5728\u65e0 TensorRT \u4f18\u5316\u7684\u60c5\u51b5\u4e0b&#xff0c;\u5b9e\u9645\u541e\u5410\u6bd4\u9884\u671f\u4f4e 25% \u4ee5\u4e0a&#xff0c;\u5e76<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[103,50,43],"topic":[],"class_list":["post-105774","post","type-post","status-publish","format-standard","hentry","category-server","tag-ubuntu","tag-50","tag-43"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v20.3 - https:\/\/yoast.com\/wordpress\/plugins\/seo\/ -->\n<title>\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/105774.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u6211\u6709\u4e00\u53f0\u57fa\u4e8e Ubuntu 22.04 \u7684\u591a\u663e\u5361\u670d\u52a1\u5668&#xff0c;\u7528\u4e8e\u670d\u52a1\u5b9e\u65f6\u63a8\u7406\u8bf7\u6c42\u3002\u4e1a\u52a1\u573a\u666f\u5305\u62ec\u5927\u89c4\u6a21\u56fe\u50cf\u5206\u7c7b\u3001\u5bf9\u8c61\u68c0\u6d4b\u4e0e LLM&#xff08;\u5927\u8bed\u8a00\u6a21\u578b&#xff09;\u63a8\u7406&#xff0c;\u8fd9\u4e9b\u6a21\u578b\u5bf9\u5ef6\u8fdf\u548c\u541e\u5410\u6709\u6781\u9ad8\u8981\u6c42\u3002 \u521d\u671f&#xff0c;A5\u6570\u636e\u9047\u5230\u8fc7\u8bf8\u5982&#xff1a;\u63a8\u7406\u65f6\u5ef6\u8fdf\u6ce2\u52a8\u3001\u663e\u5b58\u5229\u7528\u4e0d\u5747\u3001\u9a71\u52a8\u4e0e\u5e93\u7248\u672c\u51b2\u7a81\u3001\u4ee5\u53ca\u591a GPU \u62a2\u5360\u5bfc\u81f4\u7684\u670d\u52a1\u7a33\u5b9a\u6027\u95ee\u9898\u3002\u5728\u65e0 TensorRT \u4f18\u5316\u7684\u60c5\u51b5\u4e0b&#xff0c;\u5b9e\u9645\u541e\u5410\u6bd4\u9884\u671f\u4f4e 25% \u4ee5\u4e0a&#xff0c;\u5e76\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/105774.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-09-15T15:40:40+00:00\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"2 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/105774.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/105774.html\",\"name\":\"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-09-15T15:40:40+00:00\",\"dateModified\":\"2026-09-15T15:40:40+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/105774.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/105774.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/105774.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/105774.html","og_locale":"zh_CN","og_type":"article","og_title":"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u6211\u6709\u4e00\u53f0\u57fa\u4e8e Ubuntu 22.04 \u7684\u591a\u663e\u5361\u670d\u52a1\u5668&#xff0c;\u7528\u4e8e\u670d\u52a1\u5b9e\u65f6\u63a8\u7406\u8bf7\u6c42\u3002\u4e1a\u52a1\u573a\u666f\u5305\u62ec\u5927\u89c4\u6a21\u56fe\u50cf\u5206\u7c7b\u3001\u5bf9\u8c61\u68c0\u6d4b\u4e0e LLM&#xff08;\u5927\u8bed\u8a00\u6a21\u578b&#xff09;\u63a8\u7406&#xff0c;\u8fd9\u4e9b\u6a21\u578b\u5bf9\u5ef6\u8fdf\u548c\u541e\u5410\u6709\u6781\u9ad8\u8981\u6c42\u3002 \u521d\u671f&#xff0c;A5\u6570\u636e\u9047\u5230\u8fc7\u8bf8\u5982&#xff1a;\u63a8\u7406\u65f6\u5ef6\u8fdf\u6ce2\u52a8\u3001\u663e\u5b58\u5229\u7528\u4e0d\u5747\u3001\u9a71\u52a8\u4e0e\u5e93\u7248\u672c\u51b2\u7a81\u3001\u4ee5\u53ca\u591a GPU \u62a2\u5360\u5bfc\u81f4\u7684\u670d\u52a1\u7a33\u5b9a\u6027\u95ee\u9898\u3002\u5728\u65e0 TensorRT \u4f18\u5316\u7684\u60c5\u51b5\u4e0b&#xff0c;\u5b9e\u9645\u541e\u5410\u6bd4\u9884\u671f\u4f4e 25% \u4ee5\u4e0a&#xff0c;\u5e76","og_url":"https:\/\/www.wsisp.com\/helps\/105774.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-09-15T15:40:40+00:00","author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"2 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/105774.html","url":"https:\/\/www.wsisp.com\/helps\/105774.html","name":"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-09-15T15:40:40+00:00","dateModified":"2026-09-15T15:40:40+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/105774.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/105774.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/105774.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"\u5982\u4f55\u5728 Ubuntu 22.04 \u4e0a\u5229\u7528 CUDA \u4e0e TensorRT \u52a0\u901f AI \u63a8\u7406\uff0c\u4f18\u5316\u591a\u663e\u5361\u670d\u52a1\u5668\u7684\u6027\u80fd\u4e0e\u7a33\u5b9a\u6027\uff1f"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/105774","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=105774"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/105774\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=105774"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=105774"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=105774"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=105774"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}