Local setup
This course page was updated until March 2022 when I left Durham University. For future updates, please visit the new version of the course pages.

Running the exercises on your own hardware #

Although you don’t need to, you might find it useful or convenient to run the exercises on your own computer. To do so, you’ll need to obtain appropriate compilers and MPI implementations.

For more general information on getting a development toolchain setup, the Faculty of Natural Sciences at Imperial College have prepared some useful guides

MacOS #

Open source tools #

I recommend homebrew for installation of packages on MacOS. You will need an MPI implementation, I recommend mpich, which can be obtained with

$ brew install mpich

Apple’s C compiler does not come with OpenMP, so you’ll need to use gcc. You can get it with

$ brew install gcc

Since MacOS ships with its own compiler toolchain, homebrew doesn’t install these compilers with clashing names, so to use this version of GCC you’ll need to run gcc-11 (rather than just gcc).

Intel toolchain #

The Intel compiler and profiling toolchain is available for free for student use. You’ll want the “Parallel Studio XE”. Note that the license under which you can use these tools is slightly constrained: read it carefully.

Linux-based systems #

Open source tools #

Most linux distributions come with MPI libraries and compilers. You may need to install appropriate packages using the package manager. On Debian-based systems you’ll need

build-essential
libmpich-dev
mpich
gcc

Intel toolchain #

As on MacOS, you can install a free version of the Intel compiler toolchain as a student.

Windows #

If you have Windows 10, you might be able to to use WSL and then follow the instructions for Linux-based systems. See also this tutorial on enabling the bash shell in Windows 10. The Intel toolchain is available on Windows, but I have no experience of it.