Treffer: 通过包络面重构的大规模粒子并行绘制算法.

A parallel rendering algorithm based on wrapping surface reconstruction was proposed for large-scale particles in distributed environments so as to visualize the particles in high quality. In the algorithm, particle clusters were represented and then rendered in the form of a series of continuous surfaces, where the distribution of the physical variable was also shown. The algorithm was parallelized in distributed environments, thus more than a hundred million particles can be visualized using a lot of processing cores. In terms of algorithm implementation, the issue of inter-block cracks during parallel computation was be solved, and the method for rapidly finding adjacent particles was presented. Meanwhile, based on visibility culling, the particle data was filtered and thus the rendering efficiency was improved. As a result, smooth surfaces with lighting can be used to expressively exhibit inner structures and physical variable distributions of particle clusters for large-scale particles. Experiment results demonstrate that using the proposed algorithm, the rendering of more than 100 million particles is realized in 5 seconds on 512 processing cores with about 60% parallel efficiency. The proposed algorithm has been successfully applied to practical simulation applications such as massively parallel non-equilibrium molecular dynamics simulations. [ABSTRACT FROM AUTHOR]

针对大规模粒子高表现可视化需求, 提出基于包络面重构的大规模粒子并行绘制算法。该算法以连续曲面的形式表示, 绘制大规模粒子的团簇表面及其物理量分布。对算法进行了分布式并行化, 从而可以通过大规模并行来处理亿以上规模的粒子数据。在算法实现上, 还解决了并行计算时的块间裂缝问题, 并提出了快速查找邻域粒子的方法, 同时, 基于可见性对粒子数据进行剔除, 提高了绘制效率。由此, 可以通过带光照效果的光滑曲面来高表现展示大规模粒子数据中的团簇结构及其物理量分布。实验结果表明, 该算法在 512 核上可在 5 s内完成上亿粒子的绘制, 并行效率可达 60%。该算法已成功应用到大规模并行非平衡分子动力学模拟等实际模拟应用中。 [ABSTRACT FROM AUTHOR]

Copyright of Journal of National University of Defense Technology / Guofang Keji Daxue Xuebao is the property of NUDT Press and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)