Mxnet环境安装

• 用anaconda作为虚拟环境的管理工具，pycharm在建立工程时，调用anaconda的虚拟环境即可。
• 在一个虚拟环境安装过cuda和cudnn之后,
  在另一个虚拟环境中安装cuda和cudnn方法:
• 首先创建Mxnet的虚拟环境:

conda create -n mxnet python==3.5

• 若删除环境,执行

conda remove -n mxnet --all

• 安装cuda

(mxnet) yuyang@oceanshadow$ conda install cudatoolkit=9.0
Collecting package metadata: done
Solving environment: done

## Package Plan ##

  environment location: /home/yuyang/anaconda3/envs/mxnet

  added / updated specs:
    - cudatoolkit=9.0


The following NEW packages will be INSTALLED:

  cudatoolkit        pkgs/main/linux-64::cudatoolkit-9.0-h13b8566_0


Proceed ([y]/n)? y

Preparing transaction: done
Verifying transaction: done
Executing transaction: done

• 安装cudnn:

(mxnet) yuyang@oceanshadow:$ conda install cudnn
Collecting package metadata: done
Solving environment: done

## Package Plan ##

  environment location: /home/yuyang/anaconda3/envs/mxnet

  added / updated specs:
    - cudnn


The following NEW packages will be INSTALLED:

  cudnn              pkgs/main/linux-64::cudnn-7.3.1-cuda9.0_0


Proceed ([y]/n)? y

Preparing transaction: done
Verifying transaction: done
Executing transaction: done

• 其他所需要的包用pip安装即可。
• 在人脸识别项目Arcfac论文实现的运行环境：

matplotlib==3.0.2
mxnet==1.3.1
numpy==1.16.1
opencv-python==4.0.0.21
scikit-image==0.14.2
scikit-learn==0.20.2
scipy==1.2.1
cudatootookit==9.0
cudnn==7.3.1

• 安装mxnet
  参考官网
• 测试mxnet-gpu:

(base) yuyang@oceanshadow:~$ source activate mxnet
(mxnet) yuyang@oceanshadow:~$ python
Python 3.5.0 |Continuum Analytics, Inc.| (default, Oct 19 2015, 21:57:25) 
[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import mxnet as mx
>>> a = mx.nd.array([1,2,3]).as_in_context(mx.gpu(0))
>>> b = mx.nd.array([4,5,6]).as_in_context(mx.gpu(0))
>>> print(a+b)

[5. 7. 9.]
<NDArray 3 @gpu(0)>