cudnn和tensorrt安装教程

先在nvidia官网下载对应安装包。一、cudnn安装#1.进入下载目录 cd~/Downloads #2.解压 xz 格式压缩包 tar-xvf cudnn-linux-x86_64-8.9.7.29_cuda11-archive.tar.xz #3.进入解压后的文件夹 cd cudnn-linux-x86_64-8.9.7.29_cuda11-archive #4.复制头文件到 CUDA 目录 sudo cp include/cudnn*.h/usr/local/cuda/include #5.复制库文件到 CUDA 库目录 sudo cp lib/libcudnn*/usr/local/cuda/lib64 #6.更新系统库缓存 sudo ldconfig #7.查看安装 cat/usr/local/cuda/include/cudnn_version.h|grep CUDNN_MAJOR-A2#defineCUDNN_MAJOR8#defineCUDNN_MINOR9#defineCUDNN_PATCHLEVEL7二、tensorrt安装#1.进入下载目录假设你把文件放在~/Downloads cd~/Downloads #2.解压 TensorRT tar-zxvf TensorRT-8.6.1.6.Linux.x86_64-gnu.cuda-11.8.tar.gz #3.移动到系统目录/usr/local推荐固定路径 sudo mv TensorRT-8.6.1.6/usr/local/tensorrt-8.6#4.添加环境变量永久生效 echo export TENSORRT_DIR/usr/local/tensorrt-8.6~/.bashrc echo export LD_LIBRARY_PATH$TENSORRT_DIR/lib:$LD_LIBRARY_PATH~/.bashrc echo export C_INCLUDE_PATH$TENSORRT_DIR/include:$C_INCLUDE_PATH~/.bashrc echo export CPLUS_INCLUDE_PATH$TENSORRT_DIR/include:$CPLUS_INCLUDE_PATH~/.bashrc echo export PATH$TENSORRT_DIR/bin:$PATH~/.bashrc #5.立即生效环境变量 source~/.bashrc #6.更新系统库缓存 sudo ldconfig #7.查看安装 ls $TENSORRT_DIR/lib/libnvinfer.so.8.6.1/usr/local/tensorrt-8.6/lib/libnvinfer.so.8.6.1