二级光线跟踪的并行计算  

Secondary ray tracing in parallel

作　　者：杨鑫[1,2] 许端清[1] 赵磊[1] 杨冰[1] 

机构地区：[1]浙江大学计算机学院,浙江杭州310027 [2]大连理工大学计算机学院,辽宁大连116023

出　　处：《浙江大学学报（工学版）》2012年第10期1796-1802,1809,共8页Journal of Zhejiang University：Engineering Science

基　　金：国家"十一五"科技支撑计划资助项目(2007BAH11B05);浙江省科技计划资助项目(2008C13079)

摘　　要：基于图形处理器(GPU)体系架构研究二级光线跟踪技术,使其交互式地绘制出真实感极强的图像.基于二级光线的特性和GPU体系结构的特点,提出以下技术:具有高并行度的加速结构快速构造方法;二级光线包的自适应生成方法;自适应的并行光线跟踪算法;面向优化带宽使用的数据动态管理方法等.实验表明,该方法可以最大限度地发挥图形处理器强大的并行计算能力,有效地使用硬件计算资源和存储资源,从而大大地加快二级光线跟踪的并行计算.Secondary ray tracing was analyzed based on graphics process unit （GPU） in order to interactively render the photorealistic image. According to the features of secondary rays and GPU architecture, the methods were proposed as follows： a highly parallel acceleration structure construction, secondary ray Packet self-adaptive generation, adaptive ray parallel tracing, and dynamic data management for optimized bandwidth usage. Experimental results show that the approach can effectively exploit the inherent parallelism of GPU to accelerate computation, utilize the processing power and memory resources provided by GPU, finally significantly improve secondary ray tracing in parallel.

关 键 词：二级光线跟踪 加速结构 并行 自适应性 图形处理器(GPU) 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]