sudo docker run -it --gpus all -e NVIDIA_DRIVER_CAPABILITIES=compute,utility sighta_ivum:latest /bin/bash 进入镜像(新容器)
sudo docker exec -it 4f1cba5972e3 /bin/bash 启动容器ID下的命令窗口
sudo docker commit 4f1cba5972e3 new_sighta_ivum 用容器ID生成镜像文件 new
sudo docker save sighta_ftp:v1.0 > sighta_ftp.tar 存储镜像
sudo docker cp ./ e5a595cf2f35:/app/ 从宿主机拷贝文件到docker
sudo docker cp e5a595cf2f35:/app/ ./ 从docker拷贝文件到宿主机
docker run --runtime=nvidia --rm nvidia/cuda nvidia-smi 验证docker内是否能找到显卡
sudo docker run -it --name=SIGHT_IVUM_GPU -v ${PWD}/:/app tensorflow/tensorflow:1.14.0-gpu-py3 /bin/bash 拉取镜像并映射宿主机目录
sudo docker run -it --name=SIGHT_IVUM_GPU --gpus all -v ${PWD}/:/app tensorflow/tensorflow:1.14.0-gpu-py3 /bin/bash 拉取镜像-gpus all
nvidia-docker run -it --name=alg-test2 --env NVIDIA_VISIBLE_DEVICES=1 -v ${PWD}/:/app zhqx511/sighta_docker:v1.0 /bin/bash 指定显卡
sudo docker tag ccc44c14345d zhqx511/sighta_training:latest 把自己的docker镜像打个个人账户下的tag
sudo nvidia-docker push zhqx511/sighta_training:latest 把自己的docker push到个人账户下
sudo docker stop sighta_ivum_v1.0 停止当前名称的docker
sudo docker rm sighta_ivum_v1.0 删除当前名称的docker
sudo docker start sighta_ivum_v1.0 启动容器,可以用docker ps -a找到容器名称
sudo docker attach sighta_ivum_v1.0 进入容器
安装 gcc g++ make环境包:sudo apt-get install build-essential