有英伟达新卡的安装

cuda与硬件驱动兼容对比

 

 

版本兼容

https://mp.weixin.qq.com/s/2B6CCjeJ0S8hZc3MtaojTA

 

pytorch-cuda=12.1   与  https://download.pytorch.org/whl/cu121是相同的作用

GPU

pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

其中cu118是cuda的版本

阿里源 https://mirrors.aliyun.com/pytorch-wheels/cu121/

 

https://data.pyg.org/whl/ 好多的torch包

一条命令可以安装pytorch GPU 、cudatoolkit(省去了手动下载好几个G的繁琐)

conda install pytorch-cuda=<CUDA_VERSION> torchvision cudatoolkit=<CUDA_VERSION> -c pytorch

互换

 conda install pytorch-cuda=11.8 torchvision  torchaudio  -c pytorch 与pip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu118

可以互换但是推荐用conda

一条命令搞定所有的英伟达GPU环境 

conda install pytorch-cuda=11.8 torchvision torchaudio cudnn -c https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/

官网

https://download.pytorch.org/whl/

CPU

pip install numpy --pre torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/nightly/cpu