도커 이미지 생성

https://keyog.tistory.com/43
https://ko.linux-console.net/?p=7788#gsc.tab=0

아래와 같이 Dockerfile 생성

vi Dockerfile

FROM nvidia/cuda:11.8.0-base-ubuntu20.04
RUN echo 'export PATH=/usr/local/cuda-11.8/bin${PATH:+:${PATH}}' >> ~/.bashrc
RUN echo 'export LD_LIBRARY_PATH=/usr/local/cuda-11.8/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}' >> ~/.bashrc
RUN echo 'export PATH=/usr/local/cuda/bin:/$PATH' >> ~/.bashrc
RUN echo 'export LD_LIBRARY_PATH=/usr/local/cuda/lib64:${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}' >> ~/.bashrc

RUN apt update
RUN apt-get install -y python3.8 python3-pip
RUN pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

ENV NVIDIA_VISIBLE_DEVICES all
ENV NVIDIA_DRIVER_CAPABILITIES compute,utility

• 도커 이미지 빌드

docker build -t mmpose:0.1 .

컨테이너 생성

• 생성된 이미지 확인
  : mmpose:0.1 이미지가 생성되었다

docker images

• 컨테이너 실행

docker run -it --name mmpose --gpus all nvidia/cuda:11.8.0-base-ubuntu20.04 /bin/bash

docker start mmpose

• GPU 작동 확인
  : pytorch에서 잘 인식되는 것을 볼 수 있다

import torch
torch.cuda.is_available()

우분투에서 윈도우 폴더 열기
https://bbeomgeun.tistory.com/139

explorer.exe .

참조
https://iambeginnerdeveloper.tistory.com/199