Ubuntu14.04+cuda6.5+opencv2.4.9+matlab2013a+caffe Configuration Record (v)--Installation Caffe

1. Install Intel MKL

Download Intel first? Parallel Studio XE Professional Edition for C + + Linux,intel free official genuine software for students, only need to apply for it. Praise I downloaded the cpp_studio_xe_2013_sp1_update3.tgz.

1. Switch to the directory where the installation files are located: cd/home/fische

2. Unzip the installation file:tar zxvf cpp_studio_xe_2013_sp1_update3.tgz

3. modify file Read and Write permissions: sudo chomd a+x cpp_studio_xe_2013_sp1_update3–r

4. Switch to unzip file directory: CD cpp_studio_xe_2013_sp1_update3

5. To install: sudo./install_gui.sh

You will then be presented with the installed graphical interface, and when you enter the application software, the installation serial number from Intel to your email is installed.

2. Setting up the cuda6.5 and MKL environments

1. Switch to/ETC/LD.SO.CONF.D folder: cd/etc/ld.so.conf.d

2. Edit cuda.conf file sudo vim cuda.conf

Enter the following in the file:

/usr/local/cuda/lib64

/lib

3. Edit intel_mkl.conf file sudo vim intel_mkl.conf

Enter the following in the file:

/opt/intel/lib/intel64
/opt/intel/mkl/lib/intel64

4. Complete the Lib file connection operation, execute:sudo ldconfig–v 3. Install Caffe

1. Installation dependencies:sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev Libopencv-dev Libboost-all-dev Libhdf5-serial-dev

2. Edit Makefile.config

Switch to Caffe file directory:cd/home/fische/caffe-master

Copy Makefile.config.examples file:cp Makefile.config.examples Makefile.config

Edit Makefile.config file:sudo vim makefile.config

Modify the configuration file according to your own situation

Note: If the default installation folder is not selected when installing the MKL, modify the Blas_include and Blas_lib folders as appropriate.

3. Compiling Caffe

Make all

Make Test

Make Runtest

PS: Hint "src/caffe/util/math_functions.cu": Error:calling a host function ("Std::signbit") from a globalfunction (" Caffe::sgnbit_kernel ") is not allowed"
Workaround:

Modify./INCLUDE/CAFFE/UTIL/MATH_FUNCTIONS.HPP 224 Lines

Delete (note): Using Std::signbit;

Modified: Define_caffe_cpu_unary_func (sgnbit, y[i] = Signbit (X[i]));

For: Define_caffe_cpu_unary_func (sgnbit, y[i] = Std::signbit (X[i]));

From blog: http://www.cnblogs.com/liangliangdetianxia/p/3980103.html

4. Compiling MATLAB wrapper

Make Matcaffe

Reference Blog

Http://www.cnblogs.com/platero/p/3993877.html

Http://www.cnblogs.com/liangliangdetianxia/p/3980103.html

Ubuntu14.04+cuda6.5+opencv2.4.9+matlab2013a+caffe configuration Record (v)--Install Caffe