为什么安装 cuda_12.2.2_535.104.05_linux.run 之后，还是没有 nvcc 命令？

Question

我的机器是 nvidia T4 GPU + ubuntu22.04

我先通过下面的命令安装驱动

sudo apt install -y nvidia-driver-535-server

等电脑重启好了，输入 nvidia-smi 查看显卡信息

╰─➤  nvidia-smi                                                                                                                                                                                                                                                                                                                                            130 ↵
Mon Sep 18 14:30:16 2023       
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.54.03              Driver Version: 535.54.03    CUDA Version: 12.2     |
|-----------------------------------------+----------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |         Memory-Usage | GPU-Util  Compute M. |
|                                         |                      |               MIG M. |
|=========================================+======================+======================|
|   0  Tesla T4                       Off | 00000000:AF:00.0 Off |                    0 |
| N/A   47C    P0              27W /  70W |      2MiB / 15360MiB |      6%      Default |
|                                         |                      |                  N/A |
+-----------------------------------------+----------------------+----------------------+

然后在 https://developer.nvidia.com/cuda-downloads?target_os=Linux&t... 下面 CUDA Toolkit 12.2

╭─pon@T4GPU ~/Downloads  
╰─➤  sudo sh cuda_12.2.2_535.104.05_linux.run
[sudo] password for pon: 

安装之后，还是没有 nvcc

╭─pon@T4GPU ~/Downloads  
╰─➤  nvcc --version                                                                                                                                  127 ↵

zsh: command not found: nvcc
╭─pon@T4GPU ~/Downloads  
╰─➤  cd /                                                                                                                                            127 ↵
╭─pon@T4GPU /  
╰─➤  fd -a -u nvcc
/usr/share/cmake-3.22/Modules/FindCUDA/run_nvcc.cmake

我的期望是，安装这个 CUDA Toolkit 之后，就有 nvcc 命令

Answer 1

原来的方式为什么不行，我懒得搞了

我直接用 apt 命令安装了

sudo apt-get install nvidia-cuda-dev  nvidia-cuda-toolkit

安装好了之后就有了

╰─➤  nvcc --version

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

但是有点奇怪的就是 Build cuda_11.5.r11.5/compiler.30672275_0 ，意思是基于 cuda11 编译？我希望的是 cuda12, 不过问题不大

╭─pon@T4GPU /  
╰─➤  fd -a -u nvcc
/etc/nvcc.profile
/usr/share/cmake-3.22/Modules/FindCUDA/run_nvcc.cmake
/usr/bin/nvcc
/usr/lib/nvidia-cuda-toolkit/bin/nvcc.profile
/usr/lib/nvidia-cuda-toolkit/bin/nvcc