摘要
介绍了一种基于并行虚拟机结构的体绘制算法 .该算法以切片为单位来划分和组织体数据 ,既降低了通信代价 ,也保证了各子任务的数据局部性 .在任务分配时 ,维护并使用性能指数数据库 ,自适应式地确定各个子任务 ,实现了负载平衡 .使用一种异步二分方法 ,所有局部图像可以在 O(logn)时间内完成合并 .针对可视化算法在虚拟机环境中的并行化实现 ,自行设计并实现了一个基于 TCP/ IP和 Socket标准开发平台 .所提出的算法利用该平台而实现 ,系统采用客户 /服务器结构 .对系统在任务规模、虚拟机规模方面的可扩展性进行了实际测量 ,并对结果进行了分析比较 .
An algorithm for volume rendering in an environment of parallel virtual machines is presented in this paper. In order to reduce the communication cost, as well as to guarantee the locality of all subtasks, the volume data are divided and organized as a series of slices. By maintaining and employing a database of the performance index, the task subdivision algorithm produces an acceptable load balancing. An asynchronous binary method is introduced, which merges all partial images in O( log n) time. An efficient developing platform based on the TCP/IP and Socket standards is built. It helps researchers to parallelize various rendering algorithms on a virtual machine. The algorithm introduced in this paper has been implemented on this platform, exploiting the classical client/server paradigm. The scalabilities of both the task size and the number of hosts are tested. The experimental results are demonstrated and analyzed.
出处
《软件学报》
EI
CSCD
北大核心
2000年第8期1087-1093,共7页
Journal of Software
基金
清华大学骨干青年人才计划资助
关键词
体绘制
并行虚拟机
可扩展性
图像合成
Volume rendering, parallel virtual machine, load balancing, scalability.