HiFiLES / HiFiLES-solver

High Fidelity Large Eddy Simulation Solver
Other
174 stars 131 forks source link
HiFiLES dev. v0.1: High Fidelity Large Eddy Simulation

=======

At the Aerospace Computing Laboratory we believe that high-order numerical schemes have the potential to advance CFD beyond the current plateau of second-order methods and RANS turbulence modeling, ushering in new levels of accuracy and computational efficiency in turbulent flow simulations. HiFiLES (High Fidelity Large Eddy Simulation) is released as a freely available tool to unify the research community, promoting the advancement and wider adoption of high-order methods. The code is designed as an ideal base for further development on a variety of architectures.

HiFiLES is under active development in the Aerospace Computing Lab in the Department of Aeronautics and Astronautics at Stanford University and has been released under an open-source license (GNU General Public License v3.0).

HiFiLES Introduction

High-order numerical methods for flow simulations capture complex phenomena like vortices and separation regions using fewer degrees of freedom than their low-order counterparts. The High Fidelity (HiFi) provided by the schemes, combined with turbulence models for small scales and wall interactions, gives rise to a powerful Large Eddy Simulation (LES) software package. HiFiLES is an open-source, high-order, compressible flow solver for unstructured grids built from the ground up to take full advantage of parallel computing architectures. It is specially well-suited for Graphical Processing Unit (GPU) architectures. HiFiLES is written in C++. The code uses the MPI protocol to run on multiple processors, and CUDA to harness GPU performance.

HiFiLES Dev. Ver. 0.1 contains the following capabilities:

HiFiLES Installation

Installation is accomplished using the GNU build system AutoTools. The necessary commands for the basic installation (CPU, serial, no BLAS required) are as follows:

1) ./configure 2) make

The compiled executable will be located in ./bin/HiFiLES relative to this README.

To specify a BLAS library, compile for MPI, etc., additional options are available during the configuration process. The BASH shell script "configure_run.sh" contains example usage of all relevant configuration options, and can either be used as-is or as a template for a custem configuration script.

For example, to enable MPI support, simply change the PARALLEL="NO" flag to PARALLEL="YES", and change the value of MPICC (the MPI C compiler) if needed. Likewise, for linking with BLAS, simply select which type of BLAS you will be using (the basic CBLAS library and the ATLAS BLAS library are both supported for Linux users, and Accelerate BLAS is supported for Mac users), and then specify the location of the library (.a) and header (.h) files as shown in the BLAS_LIB and BLAS_HEADER variables.

For building on GPU-enabled devices, the process is very similar to enabling MPI or BLAS support. Simply specify the compiler (typically 'nvcc'), the location of header & library files, and (optionally) the architecture you are compiling for. Options for the architecture are listed in configure_run.sh.

Once the "configure_run.sh" shell script has been modified to suit your needs, simply use the following commands to install (assuming you are using a BASH-based Linux distro):

1) bash configure_run.sh 2) make

If you encounter problems with the above procedure, you can copy the supplied handwritten file makefiles/Makefile to the master directory and edit as required for your system. You will also need a file makefile.in tailored to your system - see the supplied examples in the makefiles subdirectory for guidance.

Thanks for building, and happy running!

HiFiLES Developers

Current developers:

Antony Jameson Thomas V. Jones Professor of Engineering, Department of Aeronautics & Astronautics. Stanford University.

Francisco Palacios Engineering Research Associate, Department of Aeronautics & Astronautics. Stanford University.

Jonathan Bull Postdoc Scholar, Department of Aeronautics & Astronautics. Stanford University.

Kartikey Asthana, Jacob Crabill, Thomas Economon, Manuel Lopez, David Manosalvas, Joshua Romero, Abhishek Sheshadri, Jerry Watkins Ph.D. Candidates (developers in alphabetical order), Department of Aeronautics & Astronautics. Stanford University.

Past developers:

Patrice Castonguay, Antony Jameson, Peter Vincent, David Williams.