Skip to content
dhr edited this page Sep 5, 2011 · 7 revisions

Building EVP

  1. Download prerequisites

    • Premake, which can be downloaded for many platforms at the Premake downloads page.

    • Git can be downloaded from the Git home page.

    • An OpenCL SDK appropriate for your target hardware. Mac OS X 10.6 or later includes an OpenCL SDK. On Linux or Windows, you will have to download one appropriate for the hardware you intend to run the EVP software on---either your CPU or GPU. NVIDIA's OpenCL SDK can be found at the CUDA downloads page.

  2. Download the EVP source

    The following commands will download the source into a directory named "evp-tools". Navigate to where you want this directory stored, then run:

     git clone http://github.com/dhr/evp-tools.git
     cd evp-tools
     git submodule init
     git submodule update
    
  3. Run Premake to generate build files

    Premake is capable of generating many different kinds of build files, including Makefiles, Visual Studio projects, and Xcode projects. Run premake4 --help for a full list of available options and build file targets.

    The Premake script included for EVP makes it easy to build the EVP command line tool and EVP's MATLAB bindings. For instance, to generate a Makefile to build the command line tool, navigate to the "evp-tools" directory and run

     premake4 --evp gmake
    

    To generate a Makefile for the MATLAB bindings, you must supply the path of MATLAB's root directory. This can be discovered by typing matlabroot at the MATLAB prompt. Then run

     premake4 --matlabroot="insert path to MATLAB root directory here" gmake
    
  4. Build EVP

    At this point you can use the build files generated in the previous step to build EVP. For instance, if you generated a Makefile, simply run make, and if the build succeeds, the EVP binary will be located in the "bin" directory.

Clone this wiki locally