Installation

Prerequisites

• At a bare minimum you need a version 1.5 or later JVM. If you want to use any CPU acceleration, you need an Oracle JVM. OpenJDK (the default on Linux) will not correctly load the native libraries. BIDMat will run as a pure java process without any acceleration at all, but that kills all the fun of using it.
• You need a 64-bit Linux, Windows 7 or Intel Mac to use the native CPU libraries. BIDMat may run on later or earlier 64-bit windows but is untested. Our Linux testing has been on RedHat Enterprise Linux 6 and Debian 6 without any issues. Other Linuxes are likely to work as well. We are not planning to support 32-bit OSes.
• You need an installation of Scala from http://www.scala-lang.org/. You should use version 2.9.2. Make sure that scala is on your path and can be run by typing "scala". Try running scala by itself, start it up, enter some simple math expressions (e.g. "4+5") to make sure its working.
• To use CUDA with an NVidia GPU you need to have a current version of the CUDA driver and CUDA toolkit installed from here https://developer.nvidia.com/cuda-toolkit. Check the INSTALLING.txt file in BIDMat for info on what GPUs and versions of the CUDA toolkit are supported.

Libraries

BIDMat is housed on Github at https://github.com/jcanny/BIDMat. To get it, do the following:

• If you haven't already: install git on your machine. If you're running windows, it will help to install cygwin.
• Choose a parent directory where you would like to keep the code, and run git clone https://github.com/jcanny/BIDMat.git
• This will create a directory BIDMat under the parent, containing executables, libs and source code.
• Follow the directions in BIDMat/INSTALLING.txt

If you're on Linux, running the "bidmat" script should work. In windows there is a bidmat.cmd script that does similar work. Its a good idea to look in these scripts and make sure the options (e.g. JVM heap size) are set to reasonable values for your machine.