Installation of VM2D¶
This section describes the information required to install VM2D on your computer.
Installation is possible both on computers under Linux and Windows.
Necessary software¶
Compiling VM2D requires the following software to be installed:
cmake automation system for building software from the source code,
C++ compiler supporting OpenMP technologie and the С++11 standard,
MPI libraries, for example, Open MPI or Microsoft MPI,
Eigen libraries (not necessarily, there are source codes in the “include” folder),
if there is a Nvidia GPU supporting CUDA technologie, CUDA Toolkit is required for its usage.
Paraview is useful for viewing the simulation results.
Downloading source codes¶
To install the VM2D code on your computer, you need to download the source codes. If Git is installed on your computer, just execute the command
git clone https://github.com/vortexmethods/VM2D.git VM2D
As a result, a subfolder VM2D will be created in the current folder and all files from the repository will be loaded into it, including source codes, examples of tasks to be solved, scripts, etc.
If Git is not installed on your computer, the source codes can be downloaded directly from the GitHub-page of the VM2D project
Preparing for installation¶
Preparing for compiling the source codes involves creating a folder build in the directory with the downloaded source codes, going to this folder and executing the command
cmake ..
If necessary, you should set the necessary keys for configuring the compilers used, specifying compilation options, etc. It may also require some modification of the CMakeLists.txt file containing the CMake settings and located in the project root directory.
In particular, to prepare source codes for compiling them on Windows using MS Visual Studio, you should, depending on the version, use one of the following commands (the Win64 option is required to use the ability to perform calculations on graphics cards Nvidia CUDA, for Visual Studio 2019 it is enabled by default for 64-bit systems)
cmake -G"Visual Studio 14 2015 Win64" ..
cmake -G"Visual Studio 15 2017 Win64" ..
cmake -G"Visual Studio 16 2019" ..
If you use a compiler on Windows that differs from the default one (as a rule, the MSVC compiler built into MS Visual Studio), for example, the Intel compiler, when preparing the source codes for compilation, you must specify, depending on version, key (note that Intel C++ Compiler 19 is integrated into Visual Studio 2019 only from Upd.4 version)
-T"Intel C++ Compiler <ver>"
On * Linux , if you want to use some other compiler instead of *gcc/g++, for example, Intel compiler (icpc), you should execute the command
CXX=icpc camke ..
Source code compilation¶
Compilation procedure depends on operating system used.
On Windows, as a rule, a project is created using CMake for its further opening and compilation in MS Visual Studio (see above).
On Linux, you can just execute the command
make
If compilation (on Linux) fails with an error related to the lack of necessary MPI libraries, you can try to clear the build folder, go to it, and instead of the cmake.. command, execute one of the following commands:
CXX=mpiCC cmake ..
CXX=mpiicpc cmake ..
and after its successful execution, repeat the compilation using the make command.
Computations using GPUs (Nvidia CUDA)¶
If the CUDA Toolkit <https://developer.nvidia.com/cuda-toolkit> _ is installed, during the cmake execution the program will be automatically configured to calculations using GPUs and NVidia CUDA technology.
When performing calculations on a node with several GPUs, by default all calculations will be performed on the device with the index #0. To avoid this, you need to open the source code file VM2D/src/VM2D/Gpu2D/Gpu2D.cpp, read the comments written in it and comment out the required line.
If the CUDA Toolkit is installed, but it is planned to perform calculations on CPUs without using the GPU capabilities, then in the CMakeLists.txt file you need to comment out the line
add_definitions(-DUSE_CUDA)
by adding symbol # to get
#add_definitions(-DUSE_CUDA)
