更换 conda 源为清华源

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
channels:
- defaults
show_channel_urls: true
default_channels:
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r
- https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2
custom_channels:
conda-forge: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
msys2: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
bioconda: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
menpo: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
pytorch: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
pytorch-lts: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
simpleitk: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud

安装 CUDA 和 CUDNN

CUDA 下载页面
https://developer.nvidia.com/cuda-downloads

nvcc -V查看安装是否成功

cuDNN 下载页面:https://developer.nvidia.com/rdp/cudnn-download

打开 C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.7
将 cuDNN 里的 bin、include、lib 复制到对应 bin、include、lib

Pytorch

下载 pytorch GPU 版

在此下载:
https://download.pytorch.org/whl/torch_stable.html

找到对应的 CUDA 版本和 python 版本
例如 torch-1.12.1+cu116-cp39-cp39-win_amd64.whl 对应 CUDA 1.16 python3.9
还需要对应 pytorch 官网下载 torchaudio torchvision。

安装 pytorch

进入下载目录
pip install ./torch-1.12.1+cu116-cp39-cp39-win_amd64.whl

验证

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
import torch

x = torch.rand(5, 3)
print(x)

print("torch version ", torch.__version__)

print("torch.cuda.is_available()", torch.cuda.is_available())

output:
tensor([[0.4556, 0.2505, 0.7284],
[0.4068, 0.0266, 0.3300],
[0.7225, 0.0090, 0.7773],
[0.2523, 0.9489, 0.8674],
[0.4931, 0.7107, 0.3855]])
torch version 1.12.1+cu116
torch.cuda.is_available() True