摘要
建筑信息模型(BIM)是建筑业信息化的一个热点,碰撞检测作为BIM所提供的一个核心功能,其计算复杂度一直受研究界和工业界的关注。随着建筑数据的规模日益扩大,碰撞检测的计算规模也持续增加,由于传统算法在结构上对计算资源的利用不具备可扩展性,难以通过增加处理器核数来持续获得性能的提升,这个问题成为限制BIM进一步发展的一个障碍。为此,提出一种新型算法,该算法以一种全新的大规模数据并行的方式实现碰撞检测,由此与现代众核处理器体系结构有着更好的匹配,其性能能够随着处理器核数的增加而成线性提升,达到理想的可扩展性。实验结果验证了算法的可扩展性和有效性。
Building Information Modeling （BIM） is a hotspot of informatization in construction industry. Collision detection is the key functionality provided by BIM and its computational complexity has been focused by research community and industry. With the scale of construction data expanding, the calculation scale of collision detection keeps increasing. It is difficult for traditional algo- rithms of collision detection to achieve performance improvement by increasing the quantity of processor cores because they are unable to use computing resources scalably. This problem has become an obstacle to the further development of BIM. It proposes a novel algorithm which carries out collision detection by a new method of massive data parallel in this paper. It has a better match with the modem many-cores processor architecture. Its performance can be improved by increasing the quantity of processor cores linearly and the ideal scalability is realized. The experimental results verified scalability and effectiveness of the algorithm.
出处
《微型电脑应用》
2015年第5期5-8,共4页
Microcomputer Applications
基金
鲁班软件大学合作计划项目
