Page 26 - 中国仿真学会通讯2020第1期
P. 26
computational speedups are less than 1 as the times especially when the computational results
occlusion points and the viewpoints are so close are intervisible.
that the computing stops soon and the
computational number of RSGs is less. The In the future, higher precision data will be
computational number of RSGs and computational needed for more accurate and realistic results.
time increase linearly and continually with the Decreasing the time ratio of copying memories is
increasing heights of viewpoints and targets until considerable. In addition, numerous allocated
the results are intervisible. Based on the threads, more parallelizable intervisibility
computational number of RSGs and computational methods and high⁃efficiency parallel technologies
time for one LoS, the average computational time can be used.
of single grid is 0.4μs.
References
The computational number of RSGs on
device are changeless and equal or greater than 1. CHEN Zhong⁃qi, REN Bo, ZHANG Bin, LAI Qiang,
which on host, and the average computational Modeling and Simulation of Visibility Analysis of
time of single grid is about 0.4μs. The maximum Target, Fire Control and Command Contral.35(2) :45
speedup reaches 10 times no matter for the -47, 2010.
computational time of one LoS or that of single
grid. 2. DAI Youhua, GUO Zhongyin, Effect of Spatial Visual
Pattern on Driving Behaviour in Expressway Tunnel
5.Conclusion and Prospect Section, Journal of Tongji University ( Natural
Science) . 39(9) :1307-1312, 2011.
The new intervisibility algorithm is
implemented to work out occlusion points on a 3. LI Hanrong, NIU Hongguang, JIA Juntao, LIU
long⁃distance line of sight. After dividing a line Canyou, LU Yi, Visibility Analysis Technology Based
into several segments, segments are computed in on Digital Sea, Hydrographic Surveying and Charting.
different GCS tiles to decrease the influence of the 33(4) :40-44, 2013.
earth curvature. Because of the high precision of
terrain data, intervisibility computing is 4. WANG Zhi⁃jie, YIN Xin, ZHU Liang⁃jia,
implemented in amounts of RSGs repeatedly. Improvement of Inter⁃Visibility Algorithm over
Terrain, Information Command Control System and
In this paper, GPU parallel computing is Simulation Technology. 26(2) :54-57, 2004.
taken into account to let every thread implement
computing of every RSG concurrently and 5. WANG Zhi⁃jie, QIU Xiao⁃gang, LI Ge, Design of
independently to improve the performance of Fast Inter⁃visibility Algorithms over Terrain Given in
intervisibility algorithm. RSG Model, Computer Simulation. 21( 12) :92 - 95,
2004.
The actual terrain data provided by DLR is
read into memories on host and device as 6. Luo Yuyu, Intervisibility Consistency Analysis of
preprocessing operation. Intervisibility parallel Multiresolution Terrain Data, master ' s thesis,
algorithm implemented by CUDA is run on high National University of Defense Technology,
performance GPU and the speedup reaches 10 Changsha, China, 2016.
7. Liankai Yao, Xiaohui Ji, Shuai Liu, Jiayue Yang,
Parallel Implementation and Performance Comparison
of BiCGStab for Massive Sparse Linear System of
Equations on GPU Libraries, UIC⁃ATC⁃ScalCom⁃
CBDCom⁃IoP 2015: 603-608, 2015.
23