基于回归森林的并行绘制系统的负载平衡策略  

Regression Forest Based Load Balancing Strategy for Parallel Rendering System

作　　者：罗小涛 段思羽 LUO Xiao-tao1,2, DUAN Si-yu1,2(1. College of Computer Science, Sichuan University, Chengdu 610065 ; 2. National Key Laboratory of Fundamental Science on Synthetic Vision, SiChuan University, Chengdu 61006)

机构地区：[1]四川大学计算机学院,成都610065 [2]四川大学视觉合成图形图像技术国家重点学科实验室,成都610065

出　　处：《现代计算机》2018年第5期41-45,共5页Modern Computer

摘　　要：并行绘制广泛应用于基于实时交互的娱乐系统中,各并行绘制节点间负载是否平衡直接影响绘制输出动画的稳定性。提出基于回归森林(RF)的负载调度算法,通过学习并预测负载的平衡划分方式,使得并行绘制系统在绘制过程中做到负载的平衡划分。实验证明,使用该方法进行负载调度,使得输出的动画连续且稳定,提升资源的利用率,同时也提高并行绘制系统输出动画的帧率。Parallel rendering is widely used in real-time interaction based entertainment system, if workloads between parallel rendering nodes bal-anced can impact the stability of the output animation. Proposes a new regression forest based load balancing strategy. By learning the bal- anced division of workload, parallel rendering system can achieve dynamic load balancing. Numerical experiments show that using the pre- diction output of regression forest model to dispatch workload makes the output animation stable, and improves resource utilization and the animation FPS rendered by parallel rendering system.

关 键 词：并行绘制 负载平衡 回归森林 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]