SIMD FEniCS project

Working repository for the FEniCS Google Summer of Code SIMD project. Most code can be found in the main/simd folder.

Development environment

I perform most of the development on Windows using Docker. A longer introduction can be found on my blog but for a quick start, the following sections should provide enough information.

Docker image (recommended)

This repository provides a Dockerfile with all required components. In the root folder of the repository run

docker build -t simd-work .

or alternatively get it from the quay.io repository using

docker pull quay.io/w1th0utnam3/simd-work

then start container with

docker run -v [host path to repo]/fenics_work:/local/fenics_work -it simd-work /bin/bash

For development, the submodules in this repository (Dolfin-X, FFC-X, etc.) are mounted into the container and should be installed with pip in editable mode. To allow this, some egg-info folders have to be generated in the repositories. This has to be done once after first cloning this repository and is performed by the check_install.py script:

docker run -v [host path to repo]/fenics_work:/local/fenics_work -it simd-work python3 /local/fenics_work/main/simd/check_install.py

Installing FEniCS locally (not updated)

Previously, I used a local installtion of FEniCS for development on linux and wrote an install script for this. However, I did not update it recently.

Required packages

The dolfinx Dockerfile shows which packages are required. As a single command:

sudo apt-get -y install \
    cmake \
    doxygen \
    g++ \
    gfortran \
    git \
    gmsh \
    graphviz \
    libboost-dev \
    libboost-filesystem-dev \
    libboost-iostreams-dev \
    libboost-math-dev \
    libboost-program-options-dev \
    libboost-system-dev \
    libboost-thread-dev \
    libboost-timer-dev \
    libeigen3-dev \
    libhdf5-openmpi-dev \
    liblapack-dev \
    libopenmpi-dev \
    libopenblas-dev \
    openmpi-bin \
    pkg-config \
    python3-dev \
    valgrind \
    wget \
    bash-completion

Install script

The install_fenics.py script downloads all major FEniCS components (fiat, ufl, dijisto, ffcx, dolfinx and Firedrake's tsfc, coffee, finat) from their repositories (master branch) and installs them in a python virtual environment.

Options

Currently, the script has the following options:

usage: install_fenics.py [-h] [-r REPO_DIR] [-y] [-l | -co] [-j JOBS]
                         [install_prefix]

install and set up an environment for FEniCS from git

positional arguments:
  install_prefix        FEniCS will be installed into this folder. If not
                        specified, the current folder will be used.

optional arguments:
  -h, --help            show this help message and exit
  -r REPO_DIR, --repo-dir REPO_DIR
                        Directory where source directories (repositories) will
                        be stored/are already located.
  -y, --yes             Respond with 'yes' to all confirmation messages.
  -l, --local           Don't clone any repositories, use local files.
  -co, --clone-only     Only clone the required repositories, no
                        install/build.
  -j JOBS, --jobs JOBS  Number of make jobs to issue for building DOLFIN ('-j'
                        parameter for make). Default is to use
                        'os.cpu_count()'.

The install_prefix will contain the fenics_env virtual environment, the dolfin install directory, etc.

Virtual environment

In order to run the script, the virtualenv and click python packages are required. Install it using:

pip3 install virtualenv click

Optionally, add the user python bin directory to your path in ~/.profile:

export PATH=$PATH:~/.local/bin

Activating the environment

If the installtion folder was specified as ~/fenics, the proper environment can be activated by

source ~/fenics/fenics_env/bin/activate
source ~/fenics/dolfinx/share/dolfin/dolfin.conf
export PETSC_DIR=~/fenics/petsc
export SLEPC_DIR=~/fenics/slepc

Debugging

On Ubuntu the default core file size limit is 0. Use

ulimit -c unlimited

to remove this restrictions. However this has to be rerun in every shell session. The core dumps will be located in the working directory. They can be inspected by launching

gdb <executable path> <core dump path>

followed by the bt (backtrace) command inside gdb. The command quit can be used to exit gdb.