Manual Installation LLVM/MPICH
Getting a proper LLVM/MPICH compiler built is substantially more difficult than our GCC/MPICH combination. We need to build LLVM using a modern GCC compiler (not just a minimal C++11 compliant GCC compiler). Also, the higher the version of GCC, does not always guarantee success.
Furthermore, your millage will vary depending on the operating system. Building LLVM is a finicky process. If you are unsure as to why you are building an LLVM compiler, it might be best to build the GCC/MPICH compiler stack instead.
Minimum System Requirements
In general, the following is required for MOOSE-based development:
C++11 compliant compiler (GCC 4.8.4, Clang 3.4.0 or greater)
(included in any of our redistributable packages if you choose to install one)
Memory: 16 GBs (debug builds)
Processor: 64-bit x86
Cmake 3.4 or greater will be needed for building LLVM and some of the optional packages distributed with PETSc that MOOSE requires. Unless your system is very old, one should be able to use their system's package manager (apt-get, yum, zypper, etc) to install a compatible version of Cmake. For older systems, you will need to obtain cmake source from http://www.cmake.org, and build it appropriately for your system.
A sane environment. This means having a clean, nothing but the bare minimum as far as available libraries go in your running environment. No additional LD_LIBRARY_PATHs, or other extra PATHs set. No strange UMASK settings. No odd aliases. This is such an important step, that we advise if possible, to create a separate account strictly for the use of these instructions. This document assumes an account called 'moose' has been created and is the account currently in use.
Lets try to make our environment as sane as possible, while setting up all the locations we will need.
module purge #(may fail with command not found) unset LD_LIBRARY_PATH unset CPLUS_INCLUDE_PATH unset C_INCLUDE_PATH export PACKAGES_DIR=/some/path/with/write/access export STACK_SRC=`mktemp -d /tmp/moose_stack_src.XXXXXX` umask 022
What ever terminal window you were in, when you performed the above commands, you _MUST_ use that same window, for the remainder of these instructions. If this window is closed, or the machine is rebooted, it will be necessary to perform the above commands again, before continuing any step. You will also _need_ to perform any exports in any previous steps you continued from.
Create the target installation location:
mkdir -p $PACKAGES_DIR
We need a modern C++11 capable compiler. Our minimum requirements are: GCC 4.8.4, Clang 3.4.0. This section will focus on building a GCC 7.3.0 compiler stack.
What version of GCC do we have?
gcc --version gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-4) Copyright (C) 2015 Free Software Foundation, Inc. This is free software; see the source for copying conditions. There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
If your version is less than 4.8.4, you will need to build a newer version. If your version is at or greater than 4.8.4, you have the option of skipping the GCC section.
cd $STACK_SRC curl -L -O http://mirrors.concertpass.com/gcc/releases/gcc-7.3.0/gcc-7.3.0.tar.gz tar -xf gcc-7.3.0.tar.gz -C .
Obtain GCC pre-reqs:
cd $STACK_SRC/gcc-7.3.0 ./contrib/download_prerequisites
Configure, build and install GCC:
mkdir $STACK_SRC/gcc-build cd $STACK_SRC/gcc-build ../gcc-7.3.0/configure --prefix=$PACKAGES_DIR/gcc-7.3.0 \ --disable-multilib \ --enable-languages=c,c++,fortran,jit \ --enable-checking=release \ --enable-host-shared \ --with-pic make -j # (where # is the number of cores available) make install
Any errors during configure/make will need to be investigated on your own. Every operating system I have come across has its own nuances of getting stuff built. Normally any issues are going to be solved by installing the necessary development libraries using your system package manager (apt-get, yum, zypper, etc). Hint: I would search the internet for 'how to build GCC 7.3.0 on (insert the name/version of your operating system here)'
In order to utilize our newly built GCC 7.3.0 compiler, we need to set some variables:
export PATH=$PACKAGES_DIR/gcc-7.3.0/bin:$PATH export LD_LIBRARY_PATH=$PACKAGES_DIR/gcc-7.3.0/lib64:$PACKAGES_DIR/gcc-7.3.0/lib:$PACKAGES_DIR/gcc-7.3.0/lib/gcc/x86_64-unknown-linux-gnu/7.3.0:$PACKAGES_DIR/gcc-7.3.0/libexec/gcc/x86_64-unknown-linux-gnu/7.3.0:$LD_LIBRARY_PATH
We will clone all the necessary repositories involved with building LLVM/Clang from source:
mkdir -p $STACK_SRC/llvm-src cd $STACK_SRC/llvm-src git clone https://github.com/llvm-mirror/llvm.git git clone https://github.com/llvm-mirror/clang.git $STACK_SRC/llvm-src/llvm/tools/clang git clone https://github.com/llvm-mirror/compiler-rt.git $STACK_SRC/llvm-src/llvm/projects/compiler-rt git clone https://github.com/llvm-mirror/libcxx.git $STACK_SRC/llvm-src/llvm/projects/libcxx git clone https://github.com/llvm-mirror/libcxxabi.git $STACK_SRC/llvm-src/llvm/projects/libcxxabi git clone https://github.com/llvm-mirror/openmp.git $STACK_SRC/llvm-src/llvm/projects/openmp git clone https://github.com/llvm-mirror/clang-tools-extra.git $STACK_SRC/llvm-src/llvm/tools/clang/tools/extra cd $STACK_SRC/llvm-src/llvm git checkout release_60 cd $STACK_SRC/llvm-src/llvm/tools/clang git checkout release_60 cd $STACK_SRC/llvm-src/llvm/projects/compiler-rt git checkout release_60 cd $STACK_SRC/llvm-src/llvm/projects/libcxx git checkout release_60 cd $STACK_SRC/llvm-src/llvm/projects/libcxxabi git checkout release_60 cd $STACK_SRC/llvm-src/llvm/projects/openmp git checkout release_60 cd $STACK_SRC/llvm-src/llvm/tools/clang/tools/extra git checkout release_60
And now we configure, build, and install Clang:
mkdir -p $STACK_SRC/llvm-src/build cd $STACK_SRC/llvm-src/build cmake ../llvm -G 'Unix Makefiles' \ -DCMAKE_INSTALL_PREFIX=$PACKAGES_DIR/llvm-6.0.1 \ -DCMAKE_INSTALL_RPATH:STRING=$PACKAGES_DIR/llvm-6.0.1/lib \ -DCMAKE_INSTALL_NAME_DIR:STRING=$PACKAGES_DIR/llvm-6.0.1/lib \ -DCMAKE_BUILD_WITH_INSTALL_RPATH=1 \ -DLLVM_TARGETS_TO_BUILD="X86" \ -DCMAKE_BUILD_TYPE=Release \ -DCMAKE_MACOSX_RPATH:BOOL=OFF \ -DPYTHON_EXECUTABLE=`which python2.7` \ -DCMAKE_CXX_LINK_FLAGS="-L$PACKAGES_DIR/gcc-7.3.0/lib64 -Wl,-rpath,$PACKAGES_DIR/gcc-7.3.0/lib64" \ -DGCC_INSTALL_PREFIX=$PACKAGES_DIR/gcc-7.3.0 \ -DCMAKE_CXX_COMPILER=$PACKAGES_DIR/gcc-7.3.0/bin/g++ \ -DCMAKE_C_COMPILER=$PACKAGES_DIR/gcc-7.3.0/bin/gcc make -j # (where # is the number of cores available) make install
The above configuration assumes you are using the custom version of GCC built in the previous section (note the several gcc-7.3.0 paths). If this is not the case, you will need to provide the correct paths to your current toolchain. It is also possible LLVM may build successfully if you omit the -D lines referencing gcc-7.3.0 entirely.
In order to utilize our newly built LLVM-Clang compiler, we need to export some variables:
export CC=clang export CXX=clang++ export PATH=$PACKAGES_DIR/llvm-6.0.1/bin:$PATH export LD_LIBRARY_PATH=$PACKAGES_DIR/llvm-6.0.1/lib:$LD_LIBRARY_PATH
Download MPICH 3.2
cd $STACK_SRC curl -L -O http://www.mpich.org/static/downloads/3.2/mpich-3.2.tar.gz tar -xf mpich-3.2.tar.gz -C .
Now we create an out-of-tree build location, configure, build, and install it
mkdir $STACK_SRC/mpich-3.2/llvm-build cd $STACK_SRC/mpich-3.2/llvm-build ../configure --prefix=$PACKAGES_DIR/mpich-3.2 \ --enable-shared \ --enable-sharedlibs=clang \ --enable-fast=O2 \ --enable-debuginfo \ --enable-totalview \ --enable-two-level-namespace \ FC=gfortran \ F77=gfortran \ F90='' \ CFLAGS='' \ CXXFLAGS='' \ FFLAGS='' \ FCFLAGS='' \ F90FLAGS='' \ F77FLAGS='' make -j # (where # is the number of cores available) make install
In order to utilize our newly built MPI wrapper, we need to set some variables:
export PATH=$PACKAGES_DIR/mpich-3.2/bin:$PATH export CC=mpicc export CXX=mpicxx export FC=mpif90 export F90=mpif90 export C_INCLUDE_PATH=$PACKAGES_DIR/mpich-3.2/include:$C_INCLUDE_PATH export CPLUS_INCLUDE_PATH=$PACKAGES_DIR/mpich-3.2/include:$CPLUS_INCLUDE_PATH export FPATH=$PACKAGES_DIR/mpich-3.2/include:$FPATH export MANPATH=$PACKAGES_DIR/mpich-3.2/share/man:$MANPATH export LD_LIBRARY_PATH=$PACKAGES_DIR/mpich-3.2/lib:$LD_LIBRARY_PATH
Download PETSc 3.9.4
cd $STACK_SRC curl -L -O http://ftp.mcs.anl.gov/pub/petsc/release-snapshots/petsc-3.9.4.tar.gz tar -xf petsc-3.9.4.tar.gz -C .
Now we configure, build, and install it
cd $STACK_SRC/petsc-3.9.4 ./configure \ --prefix=$PACKAGES_DIR/petsc-3.9.4 \ --download-hypre=1 \ --with-ssl=0 \ --with-debugging=no \ --with-pic=1 \ --with-shared-libraries=1 \ --with-cc=mpicc \ --with-cxx=mpicxx \ --with-fc=mpif90 \ --download-fblaslapack=1 \ --download-metis=1 \ --download-parmetis=1 \ --download-superlu_dist=1 \ --download-mumps=1 \ --download-scalapack=1 \ --CC=mpicc --CXX=mpicxx --FC=mpif90 --F77=mpif77 --F90=mpif90 \ --CFLAGS='-fPIC -fopenmp' \ --CXXFLAGS='-fPIC -fopenmp' \ --FFLAGS='-fPIC -fopenmp' \ --FCFLAGS='-fPIC -fopenmp' \ --F90FLAGS='-fPIC -fopenmp' \ --F77FLAGS='-fPIC -fopenmp' \ PETSC_DIR=`pwd`
Once configure is done, we build PETSc
make PETSC_DIR=$STACK_SRC/petsc-3.9.4 PETSC_ARCH=arch-linux2-c-opt all
Everything good so far? PETSc should be asking to run more make commands
make PETSC_DIR=$STACK_SRC/petsc-3.9.4 PETSC_ARCH=arch-linux2-c-opt install
And now after the install, we can run some built-in tests
make PETSC_DIR=$PACKAGES_DIR/petsc-3.9.4 PETSC_ARCH="" test
Running the tests should produce some output like the following:
[moose@centos-7 petsc-3.9.4]$ make PETSC_DIR=$PACKAGES_DIR/petsc-3.9.4 PETSC_ARCH="" test Running test examples to verify correct installation Using PETSC_DIR=/opt/moose/petsc-3.9.4 and PETSC_ARCH= C/C++ example src/snes/examples/tutorials/ex19 run successfully with 1 MPI process C/C++ example src/snes/examples/tutorials/ex19 run successfully with 2 MPI processes Fortran example src/snes/examples/tutorials/ex5f run successfully with 1 MPI process Completed test examples =========================================
Peacock (an optional MOOSE GUI frontend) uses many libraries. The easiest way to obtain these libraries, is to install miniconda, along with several miniconda/pip packages.
cd $STACK_SRC curl -L -O https://repo.continuum.io/miniconda/Miniconda2-latest-Linux-x86_64.sh sh Miniconda2-latest-Linux-x86_64.sh -b -p $PACKAGES_DIR/miniconda PATH=$PACKAGES_DIR/miniconda/bin:$PATH conda config --set ssl_verify false PATH=$PACKAGES_DIR/miniconda/bin:$PATH conda install -c idaholab python=2.7 coverage \ reportlab \ mako \ numpy \ scipy \ scikit-learn \ h5py \ hdf5 \ scikit-image \ requests \ vtk=7.1.0 \ pyyaml \ matplotlib \ pip \ lxml \ pyflakes \ pandas \ conda-build \ mock \ yaml \ pyqt \ swig --yes
Peacock (as well as the TestHarness sytem in MOOSE), does not work with Python3. Please chose Miniconda2 for Python 2.7 instead.
Next, we need to use
pip to install additional libraries not supplied by conda:
PATH=$PACKAGES_DIR/miniconda/bin:$PATH pip install --no-cache-dir pybtex livereload==2.5.1 daemonlite pylint==1.6.5 lxml pylatexenc anytree
Now that everything has been installed, its time to wrap all these environment variables up, and throw them in a bash shell profile somewhere.
Append the following contents into a new file called
#!/bin/bash ### MOOSE Environment Profile # GCC 7.3.0 # LLVM 6.0.1 # MPICH 3.2 # PETSc 3.9.4 export PACKAGES_DIR=<what ever you exported initially during the Environment setup> export PATH=$PACKAGES_DIR/llvm-6.0.1/bin:$PACKAGES_DIR/gcc-7.3.0/bin:$PACKAGES_DIR/mpich-3.2/bin:$PACKAGES_DIR/miniconda/bin:$PATH export LD_LIBRARY_PATH=$PACKAGES_DIR/llvm-6.0.1/lib:$PACKAGES_DIR/gcc-7.3.0/lib64:$PACKAGES_DIR/gcc-7.3.0/lib:$PACKAGES_DIR/gcc-7.3.0/lib/gcc/x86_64-unknown-linux-gnu/7.3.0:$PACKAGES_DIR/gcc-7.3.0/libexec/gcc/x86_64-unknown-linux-gnu/7.3.0:$PACKAGES_DIR/mpich-3.2/lib:$LD_LIBRARY_PATH export C_INCLUDE_PATH=$PACKAGES_DIR/mpich-3.2/include:$C_INCLUDE_PATH export CPLUS_INCLUDE_PATH=$PACKAGES_DIR/mpich-3.2/include:$CPLUS_INCLUDE_PATH export FPATH=$PACKAGES_DIR/mpich-3.2/include:$FPATH export MANPATH=$PACKAGES_DIR/mpich-3.2/share/man:$MANPATH export PETSC_DIR=$PACKAGES_DIR/petsc-3.9.4 export CC=mpicc export CXX=mpicxx export FC=mpif90 export F90=mpif90
Thats it! Now you can either source this file manually each time you need to work on a MOOSE based application:
Or you can permanently have it loaded each time you open a terminal by adding the above
source command in your ~/.bash_profile (or ~/.bashrc which ever your system uses).
Whith everything finished, it is now safe to remove the temporary directory containing the source tree:
if [ -d "$STACK_SRC" ]; then rm -rf "$STACK_SRC"; fi
Compiler Stack Finished
With the compiler stack ready, you can proceed to Obtaining and Building MOOSE.