Ubuntu2204LTS-x86_64安装Cuda12.4

安装Cuda之前需要保证NVIDIA驱动安装好。Cuda安装可参考NVIDIA官方文档：cuda-installation-guide-linux

#服务器拿到手时，发现已经安装了Cuda11.5.119
###如果要使用apt安装nvcc（此方式在Ubuntu22.04LTS上安装的是Cuda11.5.119）：sudo apt install nvidia-cuda-toolkit
#查看现有nvcc版本
deepseek@deepseek1:~/installPkgs$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2021 NVIDIA Corporation
Built on Thu_Nov_18_09:45:30_PST_2021
Cuda compilation tools, release 11.5, V11.5.119
Build cuda_11.5.r11.5/compiler.30672275_0

###卸载现有的Cuda11.5.119：
deepseek@deepseek1:~/installPkgs$ sudo apt remove nvidia-cuda-toolkit

我想改成使用Cuda12.4。访问NVIDIA官网cuda安装文件下载网页 ，依次选择如下选项找到下载与安装命令：

image-20250221172033572

#下载安装文件
(self-llm) deepseek@deepseek1:~/installPkgs$ wget https://developer.download.nvidia.com/compute/cuda/12.4.0/local_installers/cuda_12.4.0_550.54.14_linux.run

#执行安装
(self-llm) deepseek@deepseek1:~/installPkgs$ sudo sh cuda_12.4.0_550.54.14_linux.run
#在弹出对话中输入“accept”接受条款，然后如下选择

image-20250222162736130

安装进程完成后，输出如下内容：

===========
= Summary =
===========

Driver:   Not Selected
Toolkit:  Installed in /usr/local/cuda-12.4/

Please make sure that
 -   PATH includes /usr/local/cuda-12.4/bin
 -   LD_LIBRARY_PATH includes /usr/local/cuda-12.4/lib64, or, add /usr/local/cuda-12.4/lib64 to /etc/ld.so.conf and run ldconfig as root

To uninstall the CUDA Toolkit, run cuda-uninstaller in /usr/local/cuda-12.4/bin
***WARNING: Incomplete installation! This installation did not install the CUDA Driver. A driver of version at least 550.00 is required for CUDA 12.4 functionality to work.
To install the driver using this installer, run the following command, replacing <CudaInstaller> with the name of this run file:
    sudo <CudaInstaller>.run --silent --driver

Logfile is /var/log/cuda-installer.log

#发现系统已经创建软链接/usr/local/cuda，指向/usr/local/cuda-12.4/。后续使用此软链接
(self-llm) deepseek@deepseek1:~/installPkgs$ ls -al /usr/local/cuda
lrwxrwxrwx 1 root root 21 Feb 21 17:55 /usr/local/cuda -> /usr/local/cuda-12.4/

#按照提示修改系统环境变量PATH、LD_LIBRARY_PATH
(self-llm) deepseek@deepseek1:~/installPkgs$ sudo vi /etc/profile
###在最后添加如下内容
export PATH=$PATH:/usr/local/cuda/bin/
export LD_LIBRARY_PATH=/usr/local/cuda/lib64/

#重新打一个终端会话窗口或重新加载上述文件以启用上述两个修改或添加的系统环境变量
#重新加载上述文件的方法：source /etc/profile
#以下以重新打一个终端会话窗口为例继续后续操作
(base) deepseek@deepseek1:~$ conda activate self-llm
(self-llm) deepseek@deepseek1:~$ echo $PATH
/home/deepseek/anaconda3/envs/self-llm/bin:/home/deepseek/anaconda3/condabin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/usr/local/cuda/bin
(self-llm) deepseek@deepseek1:~$ echo $LD_LIBRARY_PATH  
/usr/local/cuda/lib64/
#可以看到当前使用的Cuda版本已经是V12.4.99
(self-llm) deepseek@deepseek1:~$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Tue_Feb_27_16:19:38_PST_2024
Cuda compilation tools, release 12.4, V12.4.99
Build cuda_12.4.r12.4/compiler.33961263_0
(self-llm) deepseek@deepseek1:~$