安装Toolkit Driver
使用 CUDA 前，要求 GPU 驱动与 cuda 的版本要匹配，匹配关系如下：
参考：https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html#cuda-major-component-versions__table-cuda-toolkit-driver-versions
输入 nvidia-smi 命令，查看 GPU 驱动版本
Thu Mar 16 18:15:53 2023
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 515.67 Driver Version: 517.00 CUDA Version: 11.7 |
|-------------------------------+----------------------+----------------------+
| 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 NVIDIA GeForce ... On | 00000000:01:00.0 On | N/A |
| N/A 56C P3 26W / N/A | 1125MiB / 8192MiB | 55% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 6465 G /Xwayland N/A |
+-----------------------------------------------------------------------------+
可以看到当前安装的驱动版本是 517.00
安装CUDA
在 Nvidia 官网选择对应版本：https://developer.nvidia.com/cuda-toolkit-archive。比如我选择的是 11.7 版本，选择 Linux ， x86_64 ， WSL-Ubuntu ， 2.0 ， deb(local) ，如下
使用提示的安装命令直接安装即可
CUDA默认安装路径为/usr/local/cuda/
安装nvcc
sudo apt install nvidia-cuda-toolkit
输入nvcc –version查看输出结果
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2019 NVIDIA Corporation
Built on Sun_Jul_28_19:07:16_PDT_2019
Cuda compilation tools, release 10.1, V10.1.243
安装cuDNN
官网CUDA 深度神经网络库 (cuDNN) | NVIDIA Developer，下载tar
tar -xvf cudnn-linux-x86_64-8.x.x.x_cudaX.Y-archive.tar.xz
sudo cp cudnn-*-archive/include/cudnn*.h /usr/local/cuda/include
sudo cp -P cudnn-*-archive/lib/libcudnn* /usr/local/cuda/lib64
sudo chmod a+r /usr/local/cuda/include/cudnn*.h /usr/local/cuda/lib64/libcudnn*
X.Y 和 v8.x.x.x 分别用你本机 CUDA 和cuDNN 的版本信息替代
安装TensorRT
访问：https://developer.nvidia.com/nvidia-tensorrt-8x-download 下载deb版本的 TensorRT
GA为稳定版本，下载稳定版本。推荐以压缩包的形式下载TensorRT，这样就可以下载多个版本的TensorRT，方便程序调用。
os="ubuntuxx04"
tag="8.x.x-cuda-x.x"
sudo dpkg -i nv-tensorrt-local-repo-${os}-${tag}_1.0-1_amd64.deb
sudo cp /var/nv-tensorrt-local-repo-${os}-${tag}/*-keyring.gpg /usr/share/keyrings/
sudo apt-get update
sudo apt-get install tensorrt
将os和tag替换成自己的版本信息
(base) root@LAPTOP-476JT8H0:~/workstation/# dpkg -l | grep TensorRT
ii libnvinfer-bin 8.6.0.12-1+cuda12.0 amd64 TensorRT binaries
ii libnvinfer-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT development libraries
ii libnvinfer-dispatch-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT development dispatch runtime libraries
ii libnvinfer-dispatch8 8.6.0.12-1+cuda12.0 amd64 TensorRT dispatch runtime library
ii libnvinfer-headers-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT development headers
ii libnvinfer-headers-plugin-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT plugin headers
ii libnvinfer-lean-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT lean runtime libraries
ii libnvinfer-lean8 8.6.0.12-1+cuda12.0 amd64 TensorRT lean runtime library
ii libnvinfer-plugin-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT plugin libraries
ii libnvinfer-plugin8 8.6.0.12-1+cuda12.0 amd64 TensorRT plugin libraries
ii libnvinfer-samples 8.6.0.12-1+cuda12.0 all TensorRT samples
ii libnvinfer-vc-plugin-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT vc-plugin library
ii libnvinfer-vc-plugin8 8.6.0.12-1+cuda12.0 amd64 TensorRT vc-plugin library
ii libnvinfer8 8.6.0.12-1+cuda12.0 amd64 TensorRT runtime libraries
ii libnvonnxparsers-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT ONNX libraries
ii libnvonnxparsers8 8.6.0.12-1+cuda12.0 amd64 TensorRT ONNX libraries
ii libnvparsers-dev 8.6.0.12-1+cuda12.0 amd64 TensorRT parsers libraries
ii libnvparsers8 8.6.0.12-1+cuda12.0 amd64 TensorRT parsers libraries
ii tensorrt 8.6.0.12-1+cuda12.0 amd64 Meta package for TensorRT
