机器学习服务器搭建

摘要
引言
主题
附录
参考

摘要

本文讲述如何搭建机器学习服务器及搭建步骤

引言

主题

安装依赖包

安装系统依赖包

修改apt源

编辑/etc/apt/sources.list文件为以下内容

    deb http://mirrors.aliyun.com/ubuntu/ xenial main restricted universe multiverse
    deb http://mirrors.aliyun.com/ubuntu/ xenial-security main restricted universe multiverse
    deb http://mirrors.aliyun.com/ubuntu/ xenial-updates main restricted universe multiverse
    deb http://mirrors.aliyun.com/ubuntu/ xenial-backports main restricted universe multiverse
    ##测试版源
    deb http://mirrors.aliyun.com/ubuntu/ xenial-proposed main restricted universe multiverse
    # 源码
    deb-src http://mirrors.aliyun.com/ubuntu/ xenial main restricted universe multiverse
    deb-src http://mirrors.aliyun.com/ubuntu/ xenial-security main restricted universe multiverse
    deb-src http://mirrors.aliyun.com/ubuntu/ xenial-updates main restricted universe multiverse
    deb-src http://mirrors.aliyun.com/ubuntu/ xenial-backports main restricted universe multiverse
    ##测试版源
    deb-src http://mirrors.aliyun.com/ubuntu/ xenial-proposed main restricted universe multiverse
    # Canonical 合作伙伴和附加
    deb http://archive.canonical.com/ubuntu/ xenial partner
    deb http://extras.ubuntu.com/ubuntu/ xenial main

执行sudo apt-get update

安装依赖文件

sudo apt-get install ssh nfs-common libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libhdf5-serial-dev protobuf-compiler --no-install-recommends libboost-all-dev libopenblas-dev liblapack-dev libatlas-base-dev libgflags-dev libgoogle-glog-dev liblmdb-dev git cmake build-essential

安装python及依赖包

安装pythonsudo apt-get install python3-pip python3-dev python3-tk
修改pip源

编辑~/.pip/pip.conf文件为以下内容

[global]
index-url = https://pypi.tuna.tsinghua.edu.cn/simple
[install]
trusted-host=mirrors.aliyun.com

安装相关python包

sudo pip3 install --upgrade pip
sudo pip3 install --upgrade protobuf
sudo pip3 install --upgrade numpy
sudo pip3 install --upgrade matplotlib
sudo pip3 install --upgrade sklearn
sudo pip3 install --upgrade scipy ipython

安装显卡1080ti驱动

方式1（手动）

删除原有驱动

sudo apt-get remove nvidia-*
sudo apt-get autoremove
sudo nvidia-uninstall

按组合键Ctrl+Alt+F1进入文本界面
停止依赖显卡的桌面服务sudo service lightdm stop
安装驱动 sudo ./NVIDIA-Linux-x86_64-3xx.xx.run -no-x-check -no-nouveau-check -no-opengl-files

NVIDIA-Linux-x86_64-3xx.xx.run 为从NVIDIA官网下载的最新驱动
重启桌面服务sudo service lightdm restart

方式2（apt-get）

添加源 sudo add-apt-repository ppa:graphics-drivers/ppa
下载新源中的包名 sudo apt-get update

安装驱动

sudo apt-get install nvidia-38*
sudo apt-get install mesa-common-dev
sudo apt-get install freeglut3-dev

安装CUDA

执行安装命令./cuda_8.0.61_375.26_linux-run

注意：安装过程中不要安装cuda自带驱动
添加环境变量

在/etc/profile文件追加export PATH=/usr/local/cuda-8.0/bin:$PATH 在/etc/ld.so.conf文件追加/usr/local/cuda-8.0/lib64 执行命令,使环境变量生效

source /etc/profile
sudo ldconfig

安装CUDNN

执行安装命令sudo dpkg -i libcudnnxxx-1+cudax.x_amd64.deb

安装TensorFlow

sudo pip3 install --upgrade tensorflow-gpu