摘要
为了充分利用GPU集群(Cluster)中各节点的资源,提高GPU集群(Cluster)整体的计算效率.本文以3G网络中海量视频质量分析为研究背景,提出了一种面向CPU和GPU集群的负载均衡策略,构建了一个GPU异构集群系统.实验表明,该负载均衡策略能很好地利用GPU和CPU计算资源、提高集群计算效率.
In order to making full use of the resource of each node of the GPU cluster, and improving the computing performance of the GPU cluster. In this article, Taking the mass video quality analysis in 3G network as research background, made a Load balancing strategy based on CPU-GPU cluster, and built a GPU heterogeneous cluster system. Through a large number of experiments show that the load balancing policy can full advantage of the GPU and CPU computing resources and improve the efficiency of cluster computing.
出处
《微电子学与计算机》
CSCD
北大核心
2012年第9期47-50,共4页
Microelectronics & Computer
基金
国家自然科学基金项目(60970012
61003031)
上海信息技术领域重点科研攻关项目(09511501000)
上海市重点学科建设项目(S30501)
关键词
GPU集群
负载均衡
计算资源
计算效率
GPU cluster
load balance
compute resource
compute efficiency
