封装,启动时需要在GPU服务器上
显卡驱动 cuda等下载需要注册账号
| 环境 | Ubuntu 24.04 LTS |
|---|---|
| 镜像 | uvicorn-gunicorn-fastapi:python3.8 |
| 显卡支持版本 | CUDA Version: 12.5 |
| cuda版本 | cuda_12.4.0_550.54.14_linux.run |
| cudnn版本 | |
| PaddlePaddle-Gpu | |
| PaddleOCR |
启动进入容器
docker run --gpus all -it tiangolo/uvicorn-gunicorn-fastapi:python3.8 --name ugf-cuda-cudnndocker exec it ugf-cuda-cudnn /bin/bash
安装cuda
1.下载cuda
CUDA Toolkit Archive | NVIDIA Developer
显卡支持cuda版本为12.5 cuda-toolkit降低一个版本12.4
wget https://developer.download.nvidia.com/compute/cuda/12.4.0/local_installers/cuda_12.4.0_550.54.14_linux.run
2.部署cuda
sudo sh cuda_12.4.0_550.54.14_linux.run
输入accept
把这里的X按空格取消掉后再进行安装
配置环境变量
echo 'export PATH=$PATH:/usr/local/cuda-12.4/bin' >>/etc/profile
echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-12.4/lib64' >>/etc/profile
echo 'export CUDA_HOME=$CUDA_HOME:/usr/local/cuda-12.4' >>/etc/profile
source /etc/profile
检查是否安装成功
如果出现下面版本号信息则安装成功
root@ef174907636d:~# nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2024 NVIDIA Corporation
Built on Tue_Feb_27_16:19:38_PST_2024
Cuda compilation tools, release 12.4, V12.4.99
Build cuda_12.4.r12.4/compiler.33961263_0
安装cudnn
1.下载cudnn
cudnn下载地址
2.安装cudnn
wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-keyring_1.1-1_all.deb
sudo dpkg -i cuda-keyring_1.1-1_all.deb
sudo apt-get update
sudo apt-get -y install cudnn
#我们这里是cuda12所以安装12
sudo apt-get -y install cudnn-cuda-12
3.查找cudnn安装位置
whereis cudnn_version.h
#返回下面
root@ef174907636d:~# whereis cudnn_version.h
cudnn_version.h: /usr/include/cudnn_version.h
4.将cudnn相关安装文件cp到cuda下
cp /usr/include/cudnn* /usr/local/cuda-12.4/include
5.检查是否安装成功
cat /usr/local/cuda-12.4/include/cudnn_version.h | grep CUDNN_MAJOR -A 2
dpkg -l |grep cudnn
安装paddlepaddle-gpu
python -m pip install paddlepaddle-gpu==2.6.2.post120 -i https://www.paddlepaddle.org.cn/packages/stable/cu120/
安装PaddleOCR
# 先卸载掉fastapi,不然会发生版本冲突
pip uninstall fastapi
pip install PaddleOCR -i https://pypi.tuna.tsinghua.edu.cn/simple
# 再安装fastapi
pip install fastapi
安装相关依赖包
apt-get install libgl1-mesa-glx
