Oleg Bessonov

Optimized parallel approach for 3D modelling of forest fire behaviour

In this paper we present methods for parallelization of 3D CFD forest fire modelling code on Non-uniform memory computers in frame of the OpenMP environment. Mathematical model is presented first. Then, some peculiarities of this class of computers are considered, along with properties and limitations of the OpenMP model. Techniques for efficient parallelization are discussed, considering different types of data processing algorithms. Finally, performance results for the parallelized algorithm are presented and analyzed (for up to 16 processors).

Parallelization of the Solution of 3D Navier-Stokes Equations for Fluid Flow in a Cavity with Moving Covers

This paper describes the numerical method of solution of 3D Navier-Stokes equations in a regular domain and direct method of parallelization of solution for distributed-memory computers. A vorticityvector -potential formulation and Finite Difference method of solution are chosen, using fractional step ADI method for vorticity equation and Fourier method for Poisson equation. Special attention is paid to singleprocessor optimization of the algorithm. Parallelization technology is given in detail, with speedup and efficiency levels achieved for 2 and 4 processors. Numerical results are presented for different geometries and Reynolds numbers.

Efficient Parallelization of the Preconditioned Conjugate Gradient Method

In this paper we present methods for efficient parallelization of the solution of pressure Poisson equation arising in 3D CFD forest fire modeling. The solution procedure employs the Conjugate Gradient method with implicit Modified ILU (MILU) preconditioner. The basic idea for parallelizing recursive incomplete-decomposition algorithms is to use a direct nested twisted approach in combination with a staircase method. Parallelization of MILU-CG solver is implemented in OpenMP environment for Non-uniform memory (NuMA) computer systems. Performance results of the parallelized algorithm are presented and analyzed for different number of processors (up to 16).

Parallel Simulation of 3D Incompressible Flows and Performance Comparison for Several MPP and Cluster Platforms

This paper describes a parallelization method for the numerical simulation of 3D incompressible viscous flow in a cylindrical domain. Implementation details are discussed for efficient parallelization on distributed memory computers with relatively slow communication links. The developed parallel code is used for the performance evaluation of several computers of different architectures, with the number of processors used from 1 to 16. The obtained results are compared to the measured computational and communication characteristics of these computers.

Convective interactions and flow stability in the czochralskimodel in the case of crystal rotation

The results of a numerical investigation of the crystal rotation effect on the flow stability are presented on a wide Prandtl number (Pr) range. For low Pr the regimes with an elevated stability threshold are determined and the mechanisms of the loss of stability, as the critical values of the Grashof number (Gr) and the rotation velocity are exceeded, are considered. For medium and high Pr the tables of flow regimes are given, the special features of stable nonaxisymmetric helical flows are considered, and the zones of partial flow stabilization are established.

Regime diagram and three-dimensional effects of convective interactions in the hydrodynamic Czochralski model

The generalized diagram of the critical Grashof numbers as functions of the Prandtl number is presented. The diagram shows the zones of occurrence of flow field and temperature fluctuations in the axisymmetric and three-dimensional formulations of the crystal growth model using the pulling from a melt. The structure of thermals at high Prandtl numbers and the distinctive features of the three-dimensional convection structure in the zones of stabilization and hazardous mode changeover at different Prandtl numbers are discussed. The effect of crystal rotation on the flow and temperature field patterns is estimated.

OpenMP parallelization of a CFD code for multicore computers: analysis and comparison

In this paper we present a parallelization method for numerical simulation of incompressible flows in regular domains on multicore computers in frame of the OpenMP programming model. The method is based on natural splitting of a computational domain for the main part of the algorithm, and on two-dimensional splitting and application of a special tridiagonal parallelization procedure for pressure Poisson equation and other implicit parts. This method is suitable for running on shared memory computer systems with non-uniform memory and demonstrates good parallelization efficiency for up to 16 threads.

THREE-DIMENSIONAL NUMERICAL SIMULATION OF THE INTERACTION BETWEEN NATURAL CONVECTION AND RADIATION IN A DIFFERENTIALLY HEATED CAVITY IN THE LOW MACH NUMBER APPROXIMATION

Many studies have been devoted to the interaction between natural convection and radiation heat transfer in a differentially heated cavity. This problem has already been treated using the Boussinesq approximation. The main purpose of t his study is to extend this interaction to the low Mach number approximation (in 3D), for both tra nsparent and participating media. The NavierStokes and energy equations written for an ideal ga s are solved using a finite volume method, while the discrete ordinates method is used to solve the radiation transfer equation. The coupling between the energy equation and the radiation transfer is d one by adding an additional source term in the energy equation and via the radiation heat exchange between the surfaces bounding the computation domain. The work is first validated using the Bouss ine q approximation mainly by investigating the distribution of the heat flux on the hot isothermal wall. Then some simulations are presented highlighting the differences between the low Mach n umber and the Boussinesq approximations.

Using a parallel CFD code for evaluation of clusters and MPPs

We consider some methods of evaluating parallel performance on clusters and MPP platforms. These methods are based on using a parallel code for the numerical simulation of 3D incompressible viscous flow in a cylindrical domain. Details of the efficient parallelization of this code are discussed for distributed memory computers with relatively slow communication links. Results of measurements of parallel performance are presented for several computers of different architectures, with the number of processors used from 1 to 16. Additionally, direct measurements of basic performance indicators are considered, such as serial performance and speed of interprocessor communications. Finally, the comparative performance analysis is presented for these two sorts of measurements.

Optimization Techniques and Performance Analysis for Different Serial and Parallel RISC-based Computers

The paper describes different methods of performance optimization of serial and parallel algorithms for modern superscalar RISC processor based computers. The limitations imposed on the performance by hierarchical organization of computer memories are discussed, followed by the comparative performance analysis of different computers and their architectural peculiarities. Finally the parallelization aspects of the solution of 3-dimensional CFD problems are considered, along with the comparison of communication characteristics of parallel computers.