ubuntu22.04安装显卡、CUDA(含多个CUDA切换)、CUDNN、pytorch

1. 「必须」更新软件列表和安装必要软件、依赖sudo apt-get update #更新软件列表 sudo apt-get install g++ sudo apt-get install gcc sudo apt-get install make

2. 禁用 nouveau 驱动ouveau是Ubuntu自带的显卡驱动，但他是核显，我这里想安装独显，就得把他禁掉1. 命令（cmd）sudo gedit /etc/modprobe.d/blacklist.conf（输入密码）2. （自动打开的）文本，在末尾# addedblacklist nouveaublacklist lbm-nouveauoptions nouveau modeset=0alias nouveau offalias lbm-nouveau offctrl+s 保存3. 更新重启sudo update-initramfs –usudo reboot # 重启电脑``````lsmod | grep nouveau # 输出内容为空，则表示成功禁用

3. 安装驱动1. 选择合适版本：官网查询，记住版本号xxx.yy2. 卸载之前的 sudo apt-get remove nvidia-* # 卸载之前的add-apt-repository ppa:graphics-drivers/ppa # 更新显卡驱动的源apt-get install nvidia-driver-xxx # xxx是显卡版本，需要修改

4. 检查nvidia-smi

5. 安装成功。CUDA version :12.4 ——最大CUDA版本

二、CUDA

CUDA-显卡 对应

1. pytorch 官网：CUDA11.8 12.1

2. 安装依赖sudo apt-get install freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev

3. 安装CUDA1. CUDA 11.82. CUDA 12.13. 一路选择下来，我最后选了runfile(local)，因为命令行少 第一行加sudo，能防止权限不够# 11.8sudo wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda_11.8.0_520.61.05_linux.runsudo sh cuda_11.8.0_520.61.05_linux.run``````# 12.1sudo wget https://developer.download.nvidia.com/compute/cuda/12.1.0/local_installers/cuda_12.1.0_530.30.02_linux.runsudo sh cuda_12.1.0_530.30.02_linux.run

4. 安装时的选择1. continue2. 写 accept3. 不要driver：在driver处按enter4. install5. （如果有多个CUDA，会问你symlink改不改）: Yes

5. 检查nvcc -V

成功！

安装CUDA失败，重装该版本（希望用不上）

环境配置之cuda的卸载（ubuntu）_ubuntu卸载cuda-CSDN博客

切换CUDA（选读）

1. 切换脚本 1. sudo vim switch-cuda.sh2. #!/usr/bin/env bash# Copyright (c) 2018 Patrick Hohenecker## Permission is hereby granted, free of charge, to any person obtaining a copy# of this software and associated documentation files (the "Software"), to deal# in the Software without restriction, including without limitation the rights# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell# copies of the Software, and to permit persons to whom the Software is# furnished to do so, subject to the following conditions:## The above copyright notice and this permission notice shall be included in all# copies or substantial portions of the Software.## THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE# SOFTWARE.# author: Patrick Hohenecker <[email protected]># version: 2018.1# date: May 15, 2018set -e# ensure that the script has been sourced rather than just executedif [[ "${BASH_SOURCE[0]}" = "${0}" ]]; then echo "Please use 'source' to execute switch-cuda.sh!" exit 1fiINSTALL_FOLDER="/usr/local" # the location to look for CUDA installations atTARGET_VERSION=${1} # the target CUDA version to switch to (if provided)# if no version to switch to has been provided, then just print all available CUDA installationsif [[ -z ${TARGET_VERSION} ]]; then echo "The following CUDA installations have been found (in '${INSTALL_FOLDER}'):" ls -l "${INSTALL_FOLDER}" | egrep -o "cuda-[0-9]+\\.[0-9]+$" | while read -r line; do echo "* ${line}" done set +e return# otherwise, check whether there is an installation of the requested CUDA versionelif [[ ! -d "${INSTALL_FOLDER}/cuda-${TARGET_VERSION}" ]]; then echo "No installation of CUDA ${TARGET_VERSION} has been found!" set +e returnfi# the path of the installation to usecuda_path="${INSTALL_FOLDER}/cuda-${TARGET_VERSION}"# filter out those CUDA entries from the PATH that are not needed anymorepath_elements=(${PATH//:/ })new_path="${cuda_path}/bin"for p in "${path_elements[@]}"; do if [[ ! ${p} =~ ^${INSTALL_FOLDER}/cuda ]]; then new_path="${new_path}:${p}" fidone# filter out those CUDA entries from the LD_LIBRARY_PATH that are not needed anymoreld_path_elements=(${LD_LIBRARY_PATH//:/ })new_ld_path="${cuda_path}/lib64:${cuda_path}/extras/CUPTI/lib64"for p in "${ld_path_elements[@]}"; do if [[ ! ${p} =~ ^${INSTALL_FOLDER}/cuda ]]; then new_ld_path="${new_ld_path}:${p}" fidone# update environment variablesexport CUDA_HOME="${cuda_path}"export CUDA_ROOT="${cuda_path}"export LD_LIBRARY_PATH="${new_ld_path}"export PATH="${new_path}"echo "Switched to CUDA ${TARGET_VERSION}."set +ereturn3. # 保存，优先选1:w !sudo tee %:wq!
2. source switch-cuda.sh # 查看拥有的CUDA版本source switch-cuda.sh XX.X # XX.X为版本号

三、CUDNN

1. 选择合适版本：官网，下载tar（或者三个包）

2. 单个包跳到3， 三个包点链接：ubuntu下的cudnn安装_cudnn安装 ubuntu-CSDN博客

3. 解压tar -xvf cudnn-linux-x86_64-8.9.6.50_cuda11-archive.tar.xz # 改文件名称# 或者可以删掉文件名，然后把想解压的文件拖到命令窗口里，自动获取文件路径。都可以

4. 进入文件夹，复制cd cudnn-linux-x86_64-8.9.6.50_cuda11-archive/sudo cp -d -r ./lib/* /usr/local/cuda-11.8/lib64/sudo cp -r ./include/* /usr/local/cuda-11.8/include/

5. 测试sudo chmod a+r /usr/local/cuda-11.8/include/cudnn.h /usr/local/cuda-11.8/lib64/libcudnn*cat /usr/local/cuda-11.8/include/cudnn_version.h | grep CUDNN_MAJOR -A 2出现的三个数字==版本号

四、pytorch

官网：Start Locally | PyTorch

1. 官网命令pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118我用的conda xxx，没有conda用pip3 也行，据说pip3快

2. 检查（依然在CMD）python # 如果报错，试python3import torchfrom torch.backends import cudnnprint(cudnn.is_available())print(torch.backends.cudnn.version())print(torch.cuda.is_available()) # True，则gpu版本的pytorch安装成功print(torch.zeros(1).cuda())如果有python编译器，如pycharm、vscode，在那里面import print 更简单，直接复制。这里是照顾没装这些编译器的人

可能报错

#error -- unsupported GNU version! gcc versions later than 8 are not supported!-CSDN博客

Gcc多版本安装和切换_安装gcc新版本,并能切换-CSDN博客

「解决」ubuntu CUDA版本什么都对，但torch.cuda.is_available()是false-CSDN博客

参考

ubuntu下，安装配置CUDA_cuda安装教程 ubuntu-CSDN博客

ubuntu20.04安装多版本cuda,切换版本_ubuntu 多版本nvcc-CSDN博客

【深度学习环境配置】ubuntu 20.04+4060 Ti+CUDA 11.8+pytorch（装机、显卡驱动、CUDA、cudnn、pytorch）_4060ti cuda cudnn-CSDN博客