Download the CMake-hdf5-1.12.1.tar.gz file from the HDF5 downloads page, CMake-based HDF5 source code, from the HDF Group website. In a terminal, enter the commands:

gunzip CMake-hdf5-1.12.1.tar.gz
tar -xf CMake-hdf5-1.12.1.tar
cd CMake-hdf5=1.12.1
ctest -S HDF5config.cmake,BUILD_GENERATOR=Unix -C Release -VV -O hdf5.log

For parallel HDF5 1.12.1, replace the ctest command line with:

ctest -S HDF5config.cmake,BUILD_GENERATOR=Unix,HPC=ON -C Release -VV -O hdf5-mpi.log

To install HDF5, copy the HDF5-1.12.1-Linux.tar.gz file to the installation directory and enter these commands:

tar -xf HDF5-1.12.1-Linux.tar.gz
export HDF5_DIR="$(pwd)/HDF5-1.12.1-Linux/HDF_Group/HDF5/1.12.1/share/cmake"

There are a variety of package managers available for macOS. The MCNP Development Team typically uses HomeBrew and/or Spack software package manager to provide such capability. As such, when a package manager is referred to in this section, it will refer to such software that provides this type of capability.

On macOS, the CMake-based version of HDF5 will only build with AppleClang, which causes problems for parallel HDF5. So, download the hdf5-1.12.1.tar.gz file from the HDF5 downloads page, Autoconf-based HDF5 source code, from the HDF5 website. Then enter:

export HDF5_ROOT=<install path>
gunzip hdf5-1.12.1.tar.gz
tar -xf hdf5-1.12.1.tar
cd hdf5-1.12.1
./configure --prefix=$HDF5_ROOT
make -j && make test && make install

For parallel HDF5, replace the configure line with:

CC=mpicc ./configure --enable-parallel --prefix=$HDF5_ROOT

CMake can also create the MCNP project solution file, mcnp.sln for use with Visual Studio. Note that compiling the MCNP code with Visual Studio will take longer than using the Ninja build tool.

Open a command prompt and create a new the MCNP build tree folder, build_VS. Go to the new folder and enter the following CMake command, depending on which version of Visual Studio is used:

Visual Studio 2019

cmake -G "Visual Studio 16 2019" ..\ -T "Intel C++ Compiler 19.2" -DCMAKE_Fortran_FLAGS_INIT=/MP

Visual Studio 2022

cmake -G "Visual Studio 17 2022" ..\ -T "Intel C++ Compiler 19.2" -DCMAKE_Fortran_FLAGS_INIT=/MP

The -DCMAKE_Fortran_FLAGS_INIT=/MP CMake flag tells the Fortran compiler to process the source code in parallel.

To open the MCNP project in Visual Studio, enter the command:

devenv mcnp.sln

Download the CMake-hdf5-1.12.1.zip file from the HDF5 downloads page, CMake-based HDF5 source code, from the HDF Group website. Build instructions depend on the version of Visual Studio used.

Visual Studio 2019

chdir <path to extracted folder>
ctest -S HDF5config.cmake,BUILD_GENERATOR=VS201964 -C Release -VV -O hdf5.log

Visual Studio 2022

164:    set (SITE_COMPILER_VERSION "11")
165:  else ()
166:    message (FATAL_ERROR "Invalid BUILD_GENERATOR must be - Unix, VS2019, VS201964, VS2017, or VS201764, VS2015, VS201564")

  elseif (BUILD_GENERATOR STREQUAL "VS202264")
    set (CTEST_CMAKE_GENERATOR "Visual Studio 17 2022")
    set (CMAKE_GENERATOR_ARCHITECTURE "x64")
    set (SITE_OS_BITS "64")
    set (SITE_COMPILER_NAME "vs2022")
    set (SITE_COMPILER_VERSION "17")

chdir <path to extracted folder>
ctest -S HDF5config.cmake,BUILD_GENERATOR=VS202264 -C Release -VV -O hdf5.log

For either version of Visual Studio, after this process is finished, extract the generated HDF5-1.12.1-win64.zip file to a location of your choice.

To build parallel HDF5, open the file HDF5options.cmake and scroll down to line 80:

79:     set (ADD_BUILD_OPTIONS "${ADD_BUILD_OPTIONS} -DHDF5_BUILD_JAVA:BOOL=OFF")
80:     set (ADD_BUILD_OPTIONS "${ADD_BUILD_OPTIONS} -DHDF5_ENABLE_THREADSAFE:BOOL=OFF")
81:  endif()

Insert this line after line 80:

    set (ADD_BUILD_OPTIONS "${ADD_BUILD_OPTIONS} -DMPI_GUESS_LIBRARY_NAME=MSMPI")

This option assures that CMake uses Microsoft MPI to build parallel HDF5

Next, replace the ctest command from above with:

ctest -S HDF5config.cmake,BUILD_GENERATOR=VS201964,MPI=ON -C Release -VV -O hdf5-mpi.log -D LOCAL_SKIP_TEST=ON

or

ctest -S HDF5config.cmake,BUILD_GENERATOR=VS202264,MPI=ON -C Release -VV -O hdf5-mpi.log -D LOCAL_SKIP_TEST=ON

depending on the Visual Studio version.

Tests typically need to be disabled on Windows machines, as some of the parallel tests tend to lock up on hard drives formatted with NTFS. If one is using a true parallel file system, that option can be omitted.

There are two ways to build on Apple Silicon: natively with Xcode command line tools using dependencies installed via homebrew or using a Docker container as described here. The homebrew path provides convenience whereas the Docker path provides access to a Fortran capable debugger. Which one you use will depend on your needs as a developer.

All terminal commands presented in this section are intended to be executed within a bash shell and should be compatible with zsh, which is the default shell under macOS.

Prerequisites

sudo mkdir -p /opt/homebrew
sudo chown -R $(whoami) /opt/homebrew
curl -L https://github.com/Homebrew/brew/tarball/master | tar xz --strip 1 -C /opt/homebrew
export PATH=/opt/homebrew/bin:$PATH

brew install cmake
brew install make
brew install gcc@14
brew install hdf5
brew install mpich

Building with Apple Clang

mkdir build_clang
cd build_clang
FC=gfortran-xx CC=clang CXX=clang++ cmake .. -DOpenMP=OFF
gmake -j

Building with GCC

mkdir build_gcc
cd build_gcc
FC=gfortran-xx CC=gcc-xx CXX=g++-xx cmake .. -DCMAKE_OSX_SYSROOT= 
gmake -j

• Open Docker Desktop settings by clicking on the gear icon at the top right of the window
• Click on Resources
• Click on Network under Resources
• Ensure that both checkboxes are selected as in Figure 1 and click Apply & restart if changes were made

• Open XQuartz settings under the XQuartz menu
• Click on Security
• Ensure that both checkboxes are selected as in Figure 2

To build an image, you have to first launch the Docker Desktop. Place the listed Dockerfile in a directory with no other files in it. Then execute the following commands in the Terminal application:

cd /path/to/Dockerfile_directory
export USERNAME=$(whoami) UID=$(id -ru)
docker build -f Dockerfile -t bash . --build-arg UID=${UID} --build-arg USERNAME=${USERNAME}

Depending on whether your MCNP_DATA directory is located within your home directory or not, execute one of the following two steps:

# Container with data directory that exists within your home directory
xhost +localhost
export USERNAME=$(whoami)
docker run --ulimit nofile=65536:65536 -v /Users/${USERNAME}:/Users/${USERNAME} --name mcnp6 -it bash

or

# Container with data directory outside your home directory
xhost +localhost
export USERNAME=$(whoami)
mkdir -p ~/MCNP_DATA
docker run --ulimit nofile=65536:65536 \
           -v /Users/${USERNAME}:/Users/${USERNAME} \
           -v /path/to/MCNP_DATA:/Users/${USERNAME}/MCNP_DATA \
           --name mcnp6 -it bash

This command maps your MCNP_DATA directory into your container. You have to ensure that /path/on/container exists otherwise the mapping will fail.

Once the container is instantiated, running it is as easy as starting the mcnp6 container and attaching to it from the Terminal as follows:

xhost +localhost; docker start mcnp6; docker attach mcnp6

If you get partway through the steps and need to start over you will have to delete the mcnp6 container and the bash image, in that order, by executing the following in the Terminal:

docker rm mcnp6; docker rmi bash

Flaky X11 Connection Between Docker and XQuartz

Too Many Files Open Error During Build During cmake Step

Here is the listing of the Dockerfile used to generate the container referenced above. The Dockerfile can be also be downloaded here.

#-*- mode:sh -*-
# A Dockerfile to enable MCNP development
#
# Author: Sriram Swaminarayan
# Date: March 14, 2025
#
# How to in three steps:
#   Build (once) - Initialize (once) - Attach (every time)
#
# 1: Build Image (once):
#   export USERNAME=$(whoami) UID=$(id -u)
#   docker build -t bash . --build-arg UID=${UID} --build-arg USERNAME=${USERNAME}
#
# 2: Instantiate container (once):
#   xhost +localhost
#   export USERNAME=$(whoami)
#   docker run --ulimit nofile=65536:65536 --name mcnp6 -v /Users/${USERNAME}:/Users/${USERNAME} -it bash
#
# 3: Attach to container (every time):
#   xhost +localhost; docker start mcnp6; docker attach mcnp6
#
# -----BEGIN DOCKER BUILD-----
# uses the latest ubuntu image
FROM ubuntu:latest

# Command line arguments
ARG UID
ARG USERNAME

# X11 forwarding to host
ENV DISPLAY="host.docker.internal:0"

CMD ["/bin/bash"]

# Create appropriate user
RUN mkdir -p /Users
RUN useradd -G sudo -s /bin/bash -u ${UID} -d /Users/${USERNAME} ${USERNAME}
RUN passwd -d ${USERNAME}

# Generic dependencies
RUN env
RUN apt-get update && \
    apt-get install -y \
            build-essential \
            cmake \
            cmake-curses-gui \
            ddd \
            emacs \
            g++-13 \
            gcc-13 \
            gdb \
            gfortran-13 \
            git \
            libhdf5-dev \
            make \
            mpich \
            net-tools \
            perl \
            python3 \
            qt6-base-dev \
            sudo \
            util-linux \
            vim \
            x11-apps 

# Switch to user and space
USER ${USERNAME}
WORKDIR /Users/${USERNAME}