本文已被:浏览 1367次 下载 2153次
Received:July 16, 2014 Revised:August 18, 2014
Received:July 16, 2014 Revised:August 18, 2014
中文摘要: 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.
Abstract: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.
keywords: CPU+GPU massive information cluster parallel
文章编号: 中图分类号: 文献标志码:
基金项目:广东省科技计划(2012A080102003);广东省省部产学研结合项目(2012B090500012)
引用文本:
罗明宇,刘其军,付燕平.CPU+GPU海量信息集群高速显示技术.计算机系统应用,2015,24(4):70-74
LUO Ming-Yu,LIU Qi-Jun,FU Yan-Ping.CPU + GPU Technologies for Massive Information Display on Cluster.COMPUTER SYSTEMS APPLICATIONS,2015,24(4):70-74
罗明宇,刘其军,付燕平.CPU+GPU海量信息集群高速显示技术.计算机系统应用,2015,24(4):70-74
LUO Ming-Yu,LIU Qi-Jun,FU Yan-Ping.CPU + GPU Technologies for Massive Information Display on Cluster.COMPUTER SYSTEMS APPLICATIONS,2015,24(4):70-74