Bending ray tracing is a technique for finding the shortest travel path from a fixed source to a fixed receiver. Ray tracing is a time-consuming computing technique in applications such as tomography, which involves a large number of source-receiver pairs. In this regard, parallel programming makes it possible to reduce the running time of a serial program significantly by breaking it into a discrete series and solve it by different processing units simultaneously. Along with the rapid development of parallel computing technologies in both hardware architecture and system software, parallel computing is growing rapidly in a broad range of scientific computing applications. In this paper, the parallelization of pseudo-bending ray tracing algorithm is presented using both task and data parallelization strategies. In the task parallelization, the bending calculation of each path section is distributed to different processors, while in the data parallelization, due to the independent calculation for each pair of source-receiver, the data parts are distributed to different processors. The performance results of the parallelizations of the pseudo-bending algorithm for ray tracing in a 3D velocity model are shown using OpenMp, which is an application programming interface for shared memory multiprocessing programming. The advantage of OpenMp programming model is its simplicity to parallelize an existing serial code. This is especially useful now that multi-core CPUs are common. The results show the effectiveness and efficiency of the approach. A significant speedup in the ray tracing implementation is achieved. This reduction in computation time allows more rays to be traced, which directly affects the accuracy of tomography results. Sufficient ray coverage is needed to obtain tomography images with perfect resolution.

A double-GPU code is developed to accelerate WENO schemes. The test problem is a compressible viscous flow. The convective terms are discretized using third-to ninth-order WENO schemes and the viscous terms are discretized by the standard fourth-order central scheme. The code written in CUDA programming language is developed by modifying a single-GPU code. The OpenMp library is used for parallel execution of the code on both the GPUs. Data transfer between GPUs which is the main issue in developing the code, is carried out by defining halo points for numerical grids and by using a CUDA built-in function. The code is executed on a PC equipped with two heterogeneous GPUs. The computational times of different schemes are obtained and the speedups with respect to the single-GPU code are reported for different number of grid points. Furthermore, the developed code is analyzed by CUDA profiling tools. The analyze helps to further increase the code performance.

In the present paper, a supersonic wind-tunnel is designed to maintain a flow with Mach number of 3 in a 30cm×30cm test section. An in-house CFD code is developed using the Roe scheme to simulate flow-field and detect location of normal shock in the supersonic wind-tunnel. In the Roe scheme, flow conditions at inner and outer sides of cell faces are determined using an upwind biased algorithm. The in-house CFD code has been parallelized using OpenMp to reduce the computational time. Also, an appropriate equation is derived to predict the optimum number of cores for running the program with different grid sizes. In the design process of the wind-tunnel, firstly geometry of the nozzle is specified by the method of characteristics. The flow in the nozzle and test section is simulated in the next step. Then, design parameters of the diffuser (convergence and divergence angles, area of the throat, and ratio of the exit area to the throat area) are determined by a trial and error method. Finally, an appropriate geometry is selected for the diffuser which satisfies all necessary criteria.