01、pip安装mxnet GPU版本

东方耀

01、pip安装mxnet GPU版本






mxnet官网：http://mxnet.incubator.apache.org/


apache基金会支持项目




注意：与本机的cuda版本要一致  


另外是否启用了Intel的加速  mkl-dnn  参考：http://www.ai111.vip/thread-1012-1-1.html




nvcc -V


nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2018 NVIDIA Corporation
Built on Sat_Aug_25_21:08:01_CDT_2018
Cuda compilation tools, release 10.0, V10.0.130




pip3 install mxnet-cu100




pip3 install mxnet-cu100
Looking in indexes: https://pypi.tuna.tsinghua.edu.cn/simple
Collecting mxnet-cu100
  Downloading https://pypi.tuna.tsinghua.edu.c ... nylinux1_x86_64.whl (540.1 MB)



Installing collected packages: graphviz, chardet, idna, urllib3, requests, mxnet-cu100
Successfully installed chardet-3.0.4 graphviz-0.8.4 idna-2.9 mxnet-cu100-1.5.1.post0 requests-2.23.0 urllib3-1.25.8





验证安装是否成功：http://mxnet.incubator.apache.org/get_started/validate_mxnet.html

进入python

import mxnet as mx

mx.test_utils.list_gpus()