Browsing by Author "Su, C. -C."

Now showing 1 - 4 of 4

Development of parallel direct simulation Monte Carlo method using a cut-cell Cartesian grid on a single graphics processor
(2014-09-20)
Lo, M. -C.
;
Su, C. -C.
;
Wu, J-S
;
Kuo, F. -A.
;
機械工程學系
This study developed a parallel two-dimensional direct simulation Monte Carlo (DSMC) method using a cut-cell Cartesian grid for treating geometrically complex objects using a single graphics processing unit (GPU). Transient adaptive sub-cell (TAS) and variable time-step (VTS) approaches were implemented to reduce computation time without a loss in accuracy. The proposed method was validated using two benchmarks: 2D hypersonic flow of nitrogen over a ramp and 2D hypersonic flow of argon around a cylinder using various free-stream Knudsen numbers. We also detailed the influence of TAS and VTS on computational accuracy and efficiency. Our results demonstrate the efficacy of using TAS in combination with VTS in reducing computation times by more than 10x. Compared to the throughput of a single core Intel CPU, the proposed approach using a single GPU enables a 13-35x increase in speed, which varies according to the size of the problem and type of GPU used in the simulation. Finally, the transition from regular reflection to Mach reflection for supersonic flow through a channel was simulated to demonstrate the efficacy of the proposed approach in reproducing flow fields in challenging problems. (C) 2014 Elsevier Ltd. All rights reserved.
Large-scale simulations on multiple Graphics Processing Units (GPUs) for the direct simulation Monte Carlo method
(2012-10-01)
Su, C. -C.
;
Smith, M. R.
;
Kuo, F. -A.
;
Wu, J. -S.
;
Hsieh, C. -W.
In this study, the application of the two-dimensional direct simulation Monte Carlo (DSMC) method using an MPI-CUDA parallelization paradigm on Graphics Processing Units (GPUs) clusters is presented. An all-device (i.e. GPU) computational approach is adopted where the entire computation is performed on the GPU device, leaving the CPU idle during all stages of the computation, including particle moving, indexing, particle collisions and state sampling. Communication between the GPU and host is only performed to enable multiple-GPU computation. Results show that the computational expense can be reduced by 15 and 185 times when using a single GPU and 16 GPUs respectively when compared to a single core of an Intel Xeon X5670 CPU. The demonstrated parallel efficiency is 75% when using 16 GPUs as compared to a single GPU for simulations using 30 million simulated particles. Finally, several very large-scale simulations in the near-continuum regime are employed to demonstrate the excellent capability of the current parallel DSMC method. (C) 2012 Elsevier Inc. All rights reserved.
Observations and analysis of a curved jet in the coma of comet 67P/Churyumov-Gerasimenko
(2016-04-01)
Lin, Z. -Y.
;
Lai, I. -L.
;
Su, C. -C.
;
Ip, W. -H.
;
Lees, J. -C.
Aims. We analyze the physical properties and dynamical origin of a curved jet of comet 67P/Churyumov-Gerasimenko that was observed repeatedly in several nucleus rotations starting on May 30 and persisting until early August, 2015. Methods. We simulated the motion of dust grains ejected from the nucleus surface under the influence of the gravity and viscous drag e ff ect of the expanding gas flow from the rotating nucleus. Results. The formation of the curved jet is a combination of the size of the dust particles (similar to 0.1-1 mm) and the location of the source region near the nucleus equator. This enhances the spiral feature of the collimated dust stream after the dust is accelerated to a terminal speed on the order of m s(-1).
Parallel Direct Simulation Monte Carlo Computation Using CUDA on GPUs
(2011)
Su, C. -C.
;
Hsieh, C. -W.
;
Smith, M. R.
;
Jermy, M. C.
;
Wu, J. -S.
In this study computations of the two-dimensional Direct Simulation Monte Carlo (DSMC) method using Graphics Processing Units (GPUs) are presented. An all-device (GPU) computational approach is adopted-where the entire computation is performed on the GPU device, leaving the CPU idle-which includes particle moving, indexing, collisions between particles and state sampling. The subsequent application to GPU computation requires various changes to the original DSMC method to ensure efficient performance on the GPU device. Communications between the host (CPU) and device (GPU) occur only during problem initialization and simulation conclusion when results are only copied from the device to the host. Several multi-dimensional benchmark tests are employed to demonstrate the correctness of the DSMC implementation. We demonstrate here the application of DSMC using a single-GPU, with speedups of 3 similar to 10 times as compared to a high-end Intel CPU (Intel Xeon X5472) depending upon the size and the level of rarefaction encountered in the simulation.