深度学习-实验室服务器环境配置操作指南
一、使用Anaconda配置Python环境
安装anaconda：bash Anaconda3-2019.10-Linux-x86_64.sh
实验室服务器没有挂公网，所以只能用离线方式安装，包括下面的pytorch和TensorFlow都是一样。
panchengchang@a-node03:~/envir_packages$ bash Anaconda3-2019.10-Linux-x86_64.sh
Welcome to Anaconda3 2019.10
In order to continue the installation process, please review the license
agreement.
Please, press ENTER to continue
>>>
按要求一直回车，要输yes就输yes就可以。
Do you accept the license terms? [yes|no]
[no] >>> yes
Anaconda3 will now be installed into this location:
/raid/620/panchengchang_19/anaconda3
- Press ENTER to confirm the location
- Press CTRL-C to abort the installation
- Or specify a different location below
[/raid/620/panchengchang_19/anaconda3] >>>
-按ENTER确认位置
-按CTRL-C中止安装
-或者在下面指定其他位置
直接回车就进行安装了，后面还需要再输一次yes
Preparing transaction: done
Executing transaction: done
installation finished.
Do you wish the installer to initialize Anaconda3
by running conda init? [yes|no]
[no] >>> yes
no change /raid/620/panchengchang_19/anaconda3/condabin/conda
no change /raid/620/panchengchang_19/anaconda3/bin/conda
no change /raid/620/panchengchang_19/anaconda3/bin/conda-env
no change /raid/620/panchengchang_19/anaconda3/bin/activate
no change /raid/620/panchengchang_19/anaconda3/bin/deactivate
no change /raid/620/panchengchang_19/anaconda3/etc/profile.d/conda.sh
no change /raid/620/panchengchang_19/anaconda3/etc/fish/conf.d/conda.fish
no change /raid/620/panchengchang_19/anaconda3/shell/condabin/Conda.psm1
no change /raid/620/panchengchang_19/anaconda3/shell/condabin/conda-hook.ps1
no change /raid/620/panchengchang_19/anaconda3/lib/python3.7/site-packages/xontrib/conda.xsh
no change /raid/620/panchengchang_19/anaconda3/etc/profile.d/conda.csh
modified /raid/620/panchengchang_19/.bashrc
==> For changes to take effect, close and re-open your current shell. <==
If you'd prefer that conda's base environment not be activated on startup,
set the auto_activate_base parameter to false:
conda config --set auto_activate_base false
Thank you for installing Anaconda3!
===========================================================================
Anaconda and JetBrains are working together to bring you Anaconda-powered
environments tightly integrated in the PyCharm IDE.
PyCharm for Anaconda is available at:
https://www.anaconda.com/pycharm
panchengchang@a-node03:~/envir_packages$
重启终端，输入conda或pip验证用户环境变量是否可以使用。
这样anaconda就安装成功了。
二、给conda配置国内源，提升下载速度
添加清华TUNA镜像源conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
设置搜索时显示通道地址conda config --set show_channel_urls yes
(base) panchengchang@a-node03:~$ conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
(base) panchengchang@a-node03:~$ conda config --set show_channel_urls yes
(base) panchengchang@a-node03:~$
三、使用conda创建虚拟环境
新建一个名叫pytorch，python版本为3.7的虚拟环境：conda create -n pytorch python=3.7
(base) panchengchang@a-node03:~$ conda create -n pytorch python=3.7
然后根据提示输y
Proceed ([y]/n)? y
Downloading and Extracting Packages
tk-8.6.10 | 3.0 MB | ############################################################################################################################# | 100%
readline-8.0 | 356 KB | ############################################################################################################################# | 100%
sqlite-3.33.0 | 1.1 MB | ############################################################################################################################# | 100%
wheel-0.36.2 | 33 KB | ############################################################################################################################# | 100%
python-3.7.9 | 45.3 MB | ############################################################################################################################# | 100%
setuptools-51.1.2 | 739 KB | ############################################################################################################################# | 100%
pip-20.3.3 | 1.8 MB | ############################################################################################################################# | 100%
xz-5.2.5 | 341 KB | ############################################################################################################################# | 100%
ncurses-6.2 | 817 KB | ############################################################################################################################# | 100%
ld_impl_linux-64-2.3 | 568 KB | ############################################################################################################################# | 100%
openssl-1.1.1i | 2.5 MB | ############################################################################################################################# | 100%
libedit-3.1.20191231 | 116 KB | ############################################################################################################################# | 100%
ca-certificates-2020 | 121 KB | ############################################################################################################################# | 100%
certifi-2020.12.5 | 141 KB | ############################################################################################################################# | 100%
libffi-3.3 | 50 KB | ############################################################################################################################# | 100%
Preparing transaction: done
Verifying transaction: done
Executing transaction: done
#
# To activate this environment, use
#
# $ conda activate pytorch
#
# To deactivate an active environment, use
#
# $ conda deactivate
(base) panchengchang@a-node03:~$
要激活此环境，使用conda activate pytorch
要停用活动环境，使用conda deactivate
(base) panchengchang@a-node03:~$ conda activate pytorch
(pytorch) panchengchang@a-node03:~$ conda deactivate
(base) panchengchang@a-node03:~$
四、安装pytorch（深度学习框架）
先进入虚拟环境
(base) panchengchang@a-node03:~$ conda activate pytorch
(pytorch) panchengchang@a-node03:~$
进入安装包路径下
(pytorch) panchengchang@a-node03:~$ ls
anaconda3 envir_packages
(pytorch) panchengchang@a-node03:~$ cd envir_packages/
(pytorch) panchengchang@a-node03:~/envir_packages$ ls
Anaconda3-2019.10-Linux-x86_64.sh torch-1.7.0+cu110-cp37-cp37m-linux_x86_64.whl
opencv_python-4.3.0.36-cp37-cp37m-manylinux2014_x86_64.whl torchvision-0.8.1+cu110-cp37-cp37m-linux_x86_64.whl
(pytorch) panchengchang@a-node03:~/envir_packages$ pip install torch-1.7.0+cu110-cp37-cp37m-linux_x86_64.whl
安装成功后再装另外一个
(pytorch) panchengchang@a-node03:~/envir_packages$ pip install torchvision-0.8.1+cu110-cp37-cp37m-linux_x86_64.whl
检查安装的pytorch版本是否能够使用GPU
虚拟环境下，输入python，进入python命令行
(pytorch) panchengchang@a-node03:~/envir_packages$ python
Python 3.7.9 (default, Aug 31 2020, 12:42:55)
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.cuda.is_available()
True
>>>
返回True则证明pytorch安装已完成。
五、一些常用的命令（Linux）
1.查看系统内的GPU使用情况
(pytorch) panchengchang@a-node03:~$ nvidia-smi
Fri Jan 15 22:44:22 2021
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 450.51.06 Driver Version: 450.51.06 CUDA Version: 11.0 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 A100-SXM4-40GB On | 00000000:07:00.0 Off | 0 |
| N/A 40C P0 313W / 400W | 21142MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 1 A100-SXM4-40GB On | 00000000:0F:00.0 Off | 0 |
| N/A 58C P0 344W / 400W | 40138MiB / 40537MiB | 99% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 2 A100-SXM4-40GB On | 00000000:47:00.0 Off | 0 |
| N/A 52C P0 340W / 400W | 27628MiB / 40537MiB | 100% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 3 A100-SXM4-40GB On | 00000000:4E:00.0 Off | 0 |
| N/A 53C P0 296W / 400W | 22026MiB / 40537MiB | 93% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 4 A100-SXM4-40GB On | 00000000:87:00.0 Off | 0 |
| N/A 30C P0 59W / 400W | 3MiB / 40537MiB | 0% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 5 A100-SXM4-40GB On | 00000000:90:00.0 Off | 0 |
| N/A 42C P0 71W / 400W | 3MiB / 40537MiB | 0% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 6 A100-SXM4-40GB On | 00000000:B7:00.0 Off | 0 |
| N/A 38C P0 60W / 400W | 3MiB / 40537MiB | 0% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
| 7 A100-SXM4-40GB On | 00000000:BD:00.0 Off | 0 |
| N/A 38C P0 62W / 400W | 3MiB / 40537MiB | 0% Default |
| | | Disabled |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 255273 C python 21139MiB |
| 1 N/A N/A 253640 C python 40135MiB |
| 2 N/A N/A 250479 C python 27625MiB |
| 3 N/A N/A 256275 C python 22023MiB |
+-----------------------------------------------------------------------------+
(pytorch) panchengchang@a-node03:~$
上面是静态查看，还可以动态查看：watch -n 0.1 nvidia-smi
动态查看命令中设置的0.1是指每隔0.1秒动态刷新
2.在程序中指定GPU
通常，每个服务器中具有多个GPU，GPU的编号是按照0,1,2…的顺序排列的。
在代码中加入：
import os
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
这里的代码指定了使用GPU 1，“1”代表指定的GPU块，可以根据nvidia-smi显示的结果选择合适的GPU。
3.根据pid查看进程详情
(pytorch) panchengchang@a-node03:~$ lsof -p 9347
(pytorch) panchengchang@a-node03:~$ ps -ef | grep 9347
其中9347为已知pid