351.bwaves
SPEC OMP2012 Benchmark Description

Benchmark Name

351.bwaves


Benchmark Author

Dr. Mark Kremenetsky, <mdk [at] sgi.com>
Silicon Graphics International Corp.
46600 Landing Parkway
Fremont, CA 94538, USA


Benchmark Program General Category

Computational Fluid Dynamics


Benchmark Description

351.bwaves numerically simulates blast waves in three dimensional transonic transient laminar viscous flow.

The initial configuration of the blast waves problem consists of a high pressure and density region at the center of a cubic cell of a periodic lattice, with low pressure and density elsewhere. Periodic boundary conditions are applied to the array of cubic cells forming an infinite network. Initially, the high pressure volume begins to expand in the radial direction as classical shock waves. At the same time, the expansion waves move to fill the void at the center of the cubic cell. When the expanding flow reaches the boundaries, it collides with its periodic images from other cells, thus creating a complex structure of interfering nonlinear waves. These processes create a nonlinear damped periodic system with energy being dissipated in time. Finally, the system will come to an equilibrium and steady state.

The algorithm implemented is an unfactored solver for the implicit solution of the compressible Navier-Stokes equations using the biconjugate gradient stabilized (Bi-CGstab) algorithm, which solves systems of non-symmetric linear equations iteratively.


Input Description

The input file describes the grid size, flow parameters, initial boundary condition and number of time steps. The three data sets, test, train and ref, differ only in grid size and number of time steps.

The test case is a 3-D laminar flow with Reynolds Number equal to 1E+05 and Prandtl Number equals to 0.72. In the reference data set, there are 53*64*2048 computational cells in the 1st, 2nd and 3rd dimensions, respectively. The problem is solved using by implicit time-marching scheme with a CFL number equals to 2.


Output Description

The transient nature of the flow and iterative solver makes bwaves a difficult problem to validate. In SPEC OMP2012 this has been addressed by comparing three different outputs. These are:


Programming Language

Fortran 77


Known portability issues

none


References

Mark Kremenetsky, Arthur Raefsky, Steve P. Reinhardt, "Poor Scalability of Parallel Shared Memory Model: Myth or Reality?", in Proceedings of International Conference on Computational Science'2003. pp.657~666 DOI

Last updated: June 20, 2012