在阿里云ECS实例上配置深度学习环境,可以按照以下步骤进行:
1. 选择并创建ECS实例
- 镜像选择:推荐使用 Ubuntu 20.04/22.04 或 CentOS 7/8(兼容性较好)。
- 实例规格:根据需求选择GPU实例(如
gn6i、gn7i等),若仅学习可用CPU实例。 - 系统盘:建议至少 50GB(深度学习库和数据集占用空间大)。
- 安全组:开放SSH(22端口)、Jupyter Notebook(8888端口)等必要端口。
2. 连接ECS实例
- 通过SSH连接:
ssh root@<你的ECS公网IP>
3. 安装基础依赖
Ubuntu/Debian:
apt update && apt upgrade -y
apt install -y python3-pip python3-dev git wget curl
CentOS:
yum update -y
yum install -y python3-pip python3-devel git wget curl
4. 安装NVIDIA驱动(仅GPU实例)
- 自动安装(推荐):
ubuntu-drivers autoinstall reboot - 验证驱动:
nvidia-smi
5. 安装CUDA和cuDNN
- CUDA:
- 访问 NVIDIA官网 选择版本(如CUDA 11.8)。
wget https://developer.download.nvidia.com/compute/cuda/11.8.0/local_installers/cuda_11.8.0_520.61.05_linux.run sudo sh cuda_11.8.0_520.61.05_linux.run
- 访问 NVIDIA官网 选择版本(如CUDA 11.8)。
- cuDNN:
- 下载对应版本并安装(需注册NVIDIA账号)。
6. 安装深度学习框架
PyTorch:
pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118
TensorFlow:
pip3 install tensorflow[and-cuda]
验证安装:
python3 -c "import torch; print(torch.cuda.is_available())"
python3 -c "import tensorflow as tf; print(tf.config.list_physical_devices('GPU'))"
7. 配置Jupyter Notebook(可选)
pip3 install jupyter
jupyter notebook --generate-config
jupyter notebook password # 设置密码
- 远程访问:
修改配置文件~/.jupyter/jupyter_notebook_config.py:c.NotebookApp.ip = '0.0.0.0' c.NotebookApp.open_browser = False c.NotebookApp.port = 8888 - 后台运行:
nohup jupyter notebook --allow-root &
8. 环境优化建议
- 持久化环境:使用
conda或virtualenv管理环境。 - 数据存储:挂载阿里云 NAS 或 OSS 存储数据集。
- 监控:安装
gpustat、htop等工具监控资源。
常见问题
- GPU不可用:检查驱动/CUDA版本与框架的兼容性。
- pip速度慢:更换阿里云镜像源:
pip config set global.index-url https://mirrors.aliyun.com/pypi/simple/
通过以上步骤,你可以在阿里云ECS上快速搭建深度学习环境。根据实际需求调整版本和配置,并注意定期更新驱动和库以保持兼容性。
CLOUD技术笔记