简记我在centos上安装mxnet的心路历程~ ~
步骤一：首先乖乖的按照官方文档的安装教程
git clone --recursive https://github.com/dmlc/mxnetcdmxnet;
make -j16
进行安装。结果出现error...（系统自带的g++版本为4.4.7，而mxnet要求>=4.8），进而我就鬼迷心窍的去下载安装了最新版本的g++6.0。
步骤二：更新g++版本，最好更新4.8或着4.9两个版本，因为cuda-7.5编译要求g++版本要<=4.9，（哎，这就是在玩数学不等式呐。。）
升级g++的教程如下：
1、下载g++4.8.5的源码安装包 wget ftp://ftp.gnu.org/gnu/gcc/gcc-4.8.5/gcc-4.8.5.tar.gz，
2、下载依赖包（gcc自带下载脚本）
./contrib/download_prerequisites
3、直接编译会出现error
所以可以提前执行unset LIBRARY_PATH该命令（也是可以在编译报错之后再执行。。）
4、编译安装
mkdir gcc-build-4.8.5
cd gcc-build-4.8.5
../configure --prefix=/usr
make && make install
5、通过gcc --version 、g++ --version查看版本号
降级gcc教程如下：
sudo yum remove--skip-broken gcc
sudo yum install--skip-broken gcc
yum install gcc-c++
yum-yinstallglibc-devel.i686 glibc-devel
步骤三：自定义安装
将make文件夹下面的config.mk文件copy到mxnet根目录下面，然后再config.mk文件中自定义修改。如：
#---------------------
# choice of compiler
#--------------------
export CC = gcc
export CXX = g++
export NVCC = nvcc
# whether compile with debug
DEBUG = 0
# the additional link flags you want to add
ADD_LDFLAGS = -L/usr/lib64/atlas
# the additional compile flags you want to add
#ADD_CFLAGS =
#---------------------------------------------
# matrix computation libraries for CPU/GPU
#---------------------------------------------
# whether use CUDA during compile
USE_CUDA = 1
# add the path to CUDA library to link and compile flag
# if you have already add them to environment variable, leave it as NONE
USE_CUDA_PATH = /usr/local/cuda-7.5
#USE_CUDA_PATH = NONE
# whether use CuDNN R3 library
USE_CUDNN = 1
# whether use cuda runtime compiling for writing kernels in native language (i.e. Python)
USE_NVRTC = 0
# whether use opencv during compilation
# you can disable it, however, you will not able to use
# imbin iterator
USE_OPENCV = 1
# use openmp for parallelization
USE_OPENMP = 1
# choose the version of blas you want to use
# can be: mkl, blas, atlas, openblas
# in default use atlas for linux while apple for osx
UNAME_S := $(shell uname -s)
ifeq ($(UNAME_S), Darwin)
USE_BLAS = apple
else
USE_BLAS = atlas
endif
# add path to intel library, you may need it for MKL, if you did not add the path
# to environment variable
USE_INTEL_PATH = NONE
# If use MKL, choose static link automatically to allow python wrapper
ifeq ($(USE_BLAS), mkl)
USE_STATIC_MKL = 1
else
USE_STATIC_MKL = NONE
endif
#----------------------------
# distributed computing
#----------------------------
# whether or not to enable multi-machine supporting
USE_DIST_KVSTORE = 0
# whether or not allow to read and write HDFS directly. If yes, then hadoop is
# required
USE_HDFS = 0
# path to libjvm.so. required if USE_HDFS=1
LIBJVM=$(JAVA_HOME)/jre/lib/amd64/server
# whether or not allow to read and write AWS S3 directly. If yes, then
# libcurl4-openssl-dev is required, it can be installed on Ubuntu by
# sudo apt-get install -y libcurl4-openssl-dev
USE_S3 = 0
修改好config.mk文件后，就可以运行make -j8进行编译了。
步骤四：安装python包
cd /mxnet/python
执行 python setup.py
到此就大功告成啦！
Tips：直接手动从github上下载的文件要使用git submodule init和git submodule update更新子模块。