Skip to content

Installation instructions

Bruno Golosio edited this page Jun 6, 2015 · 5 revisions

Requirements

In order to compile the software from source, you need a C++ compiler with OpenMP support (v3.0 or higher), which is freely available for most common platforms.In addition, your PATH must contain basic build tools such as make. NOTE ABOUT GPU (CUDA) VERSION: This distribution includes a mixed C++/CUDA version of the software, which is partially implemented for GPU systems. It requires CUDA libraries version >=5.5 and CUDA Toolkit version >=5.5. The CUDA installation can be done through the bash script make_cuda.sh, which uses the nvcc compiler. However, be aware that currently the CUDA version installation and usage is undocumented.

Downloading the source code

The source code can be downloaded from the ANNABELL downloads repository: http://www.neuralsystems.net/annabell/files

Extracting the source

Extract the source using any program able to extract/uncompress tar-gz archives. For instance, in Linux, Unix and Unix-like systems you can extract the source using the tar utility, by typing the command:

tar -xvzf annabell-x.x.x.tgz

where x.x.x refers to the version of the program that you are installing. This will create a new directory annabell under the current directory containing the source code for the distribution.

Building and installing the program

There are two main methods of installing ANNABELL from source: system wide installation and personal installation. The former requires root privileges on most systems. Both ways are described below. System wide installation After unpacking the distribution, go to the directory annabell and type the command:

./configure

You can then build the program by typing,

make

The installation can be performed with the command

make install

The default installation directory prefix is /usr/local. The installation directory can be changed with the --prefix option to configure. Consult the Sect. Personal installation below for instructions on installing the library in another location or changing other default compilation options. Try running ANNABELL:

annabell

Personal installation

After unpacking the distribution, go to the directory annabell and type the command:

./configure --prefix=DIRNAME

where DIRNAME is the name of the directory where you want to install the binary executable. You can then build the program by typing,

make

The installation can be performed with the command

make install

This command will install the executable in the installation directory. You should add the installation directory to the PATH environmental variable. For instance, if you use the bash shell you can add the following lines to the file .bashrc (or bashrc_private if you use it) in your home directory:

export PATH=${PATH}:DIRNAME:

If you use the tcsh shell you can add the following lines to the file .cshrc (or cshrc_private if you use it) in your home directory:

setenv PATH ${PATH}:DIRNAME:

Try running ANNABELL:

annabell

Platforms known to compile and run ANNABELL

Linux (32-bit/64-bit) In principle, it should be possible to compile and run ANNABELL in any platform that supports a C++ compiler and OpenMP version >=3.0.

Clone this wiki locally