So I needed to run some numerical simulations concerning fluid dynamics, and for this I needed to have specific versions of the Fast Fourier Transform libraries, Fortran compilers, and so.
So the author of the code for these simulations handed a virtual machine for these runs.
Nice, but virtual machines are heavy if you want to play a bit with the runs.
With a Docker environment:

  • if you want to run in parallel, the number of processor is inherited from the host (by default).
  • the input/ouput of data is smoother. This is important since you can use the tools that are already in your host to make plots during the run.
  • performance overhead is almost zero, since there’s no hypervisor in the middle.

So here is a Docker version I made from that virtual machine.

Here are some nice pictures generated by that HD/MHD (Hydro and Magneto-hydro Dynamics) fluid code.