CPU+GPU海量信息集群高速显示技术  

CPU+GPU Technologies for Massive Information Display on Cluster

作　　者：罗明宇 刘其军 付燕平 

机构地区：[1]广东粤铁瀚阳科技有限公司,广州510630

出　　处：《计算机系统应用》2015年第4期70-74,共5页Computer Systems & Applications

基　　金：广东省科技计划(2012A080102003);广东省省部产学研结合项目(2012B090500012)

摘　　要：针对集群显示系统中存在的CPU多核闲置、GPU利用不足、CPU与GPU结合困难等问题,研究了CPU多核多线程处理、GPU并行处理及CPU+GPU整合运算等技术,提出并构建了CPU+GPU集群并行显示系统,提升了集群并行显示系统的综合运算能力,实验结果表明CPU+GPU集群并行显示技术是有效的,为海量信息高速显示提供了有效的解决方案.There is much waste of computing resources in cluster, such as setting aside multi-core CPU, underutilization of GPU, separating the use of CPU and GPU. To use CPU and GPU better, some technologies, such as multicore CPU processing, GPU parallel processing and CPU ＋ GPU integrated processing, are researched in this paper. A CPU ＋ GPU cluster parallel display system is presented to improve the parallel display capability of cluster tiled system. The experimental results show that the CPU ＋ GPU cluster parallel display system presented is feasible and can greatly improve the display speed of massvie information.

关 键 词：CPU+GPU 海量 集群 并行 

分 类 号：TP332[自动化与计算机技术—计算机系统结构]